From bcec64ff89b28d9137531c7df5210a43a32fbe52 Mon Sep 17 00:00:00 2001
From: Cesaire Joris Kuete Fouodo <fouodo@imbs.uni-luebeck.de>
Date: Wed, 17 Jul 2024 14:49:49 +0200
Subject: [PATCH] Add Rmd

---
 .Rbuildignore                          |   1 +
 README.Rmd                             |  28 ++++
 README.md                              | 197 +++++--------------------
 README_files/figure-gfm/pressure-1.png | Bin 0 -> 22985 bytes
 4 files changed, 67 insertions(+), 159 deletions(-)
 create mode 100644 README.Rmd
 create mode 100644 README_files/figure-gfm/pressure-1.png

diff --git a/.Rbuildignore b/.Rbuildignore
index f42871e..8938d6b 100644
--- a/.Rbuildignore
+++ b/.Rbuildignore
@@ -2,3 +2,4 @@
 ^\.Rproj\.user$
 ^.*test_code*
 ^\.github$
+^\.README_files*
diff --git a/README.Rmd b/README.Rmd
new file mode 100644
index 0000000..02ef439
--- /dev/null
+++ b/README.Rmd
@@ -0,0 +1,28 @@
+---
+title: "fuseMLR"
+author: Cesaire J. K. Fouodo
+output: 
+  md_document:
+    variant: gfm
+    preserve_yaml: true
+---
+
+```{r setup, include=FALSE}
+knitr::opts_chunk$set(echo = TRUE)
+```
+
+<!-- badges: start -->
+  [![R-CMD-check](https://github.com/imbs-hl/fuseMLR/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/imbs-hl/fuseMLR/actions/workflows/R-CMD-check.yaml)
+  [![Lifecycle: experimental](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://lifecycle.r-lib.org/articles/stages.html#experimental)
+  [![CRAN downloads](http://www.r-pkg.org/badges/version/fuseMLR)](http://cranlogs.r-pkg.org/badges/grand-total/fuseMLR)
+  [![Stack Overflow](https://img.shields.io/badge/stackoverflow-questions-orange.svg)](https://stackoverflow.com/questions/tagged/fuseMLR)
+<!-- badges: end -->
+
+### fuseMLR
+Cesaire J. K. Fouodo
+
+### Introduction
+Recent technological advances have enabled the simultaneous targeting of multiple pathways to enhance therapies for complex diseases. This often results in the collection of numerous data entities across various layers of patient groups, posing a challenge for integrating all data into a single analysis. Ideally, patient data will overlap across layers, allowing for early or intermediate integrative techniques. However, these techniques are challenging when patient data does not overlap well. Additionally, the internal structure of each data entity may necessitate specific statistical methods rather than applying the same method across all layers. Late integration modeling addresses this by analyzing each data entity separately to obtain layer-specific results, which are then integrated using meta-analysis. Currently, no R package offers this flexibility.
+
+We introduce the fuseMLR package for late integration modeling in R. This package allows users to define studies with multiple layers, data entities, and layer-specific machine learning methods. FuseMLR is user-friendly, enabling the training of different models across layers and automatically conducting meta-analysis once layer-specific training is completed. Additionally, fuseMLR allows for variable selection at the layer level and makes predictions for new data entities.
+
diff --git a/README.md b/README.md
index 35da2e4..ffa54e4 100644
--- a/README.md
+++ b/README.md
@@ -1,171 +1,50 @@
-<!-- badges: start -->
-  [![R-CMD-check](https://github.com/imbs-hl/fuseMLR/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/imbs-hl/fuseMLR/actions/workflows/R-CMD-check.yaml)
-  [![Lifecycle: experimental](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://lifecycle.r-lib.org/articles/stages.html#experimental)
-  [![CRAN downloads](http://www.r-pkg.org/badges/version/fuseMLR)](http://cranlogs.r-pkg.org/badges/grand-total/fuseMLR)
-  [![Stack Overflow](https://img.shields.io/badge/stackoverflow-questions-orange.svg)](https://stackoverflow.com/questions/tagged/fuseMLR)
-<!-- badges: end -->
-  
-### fuseMLR
-Cesaire J. K. Fouodo
-
-### Introduction
-Recent technological advances have enabled the simultaneous targeting of multiple pathways to enhance therapies for complex diseases. This often results in the collection of numerous data entities across various layers of patient groups, posing a challenge for integrating all data into a single analysis. Ideally, patient data will overlap across layers, allowing for early or intermediate integrative techniques. However, these techniques are challenging when patient data does not overlap well. Additionally, the internal structure of each data entity may necessitate specific statistical methods rather than applying the same method across all layers. Late integration modeling addresses this by analyzing each data entity separately to obtain layer-specific results, which are then integrated using meta-analysis. Currently, no R package offers this flexibility.
-
-We introduce the fuseMLR package for late integration modeling in R. This package allows users to define studies with multiple layers, data entities, and layer-specific machine learning methods. FuseMLR is user-friendly, enabling the training of different models across layers and automatically conducting meta-analysis once layer-specific training is completed. Additionally, fuseMLR allows for variable selection at the layer level and makes predictions for new data entities.
-
-### Installation
-Installation from Github:
-```R
-devtools::install_github("imbs-hl/fuseMLR")
-```
-
-### Usage
-For usage in R, see ?fuseMLR in R. Most importantly, see the Examples section. 
-
-The provided example, utilizing simulated data, mirrors a common scenario in multi-omics analysis. It involves data collected from three distinct layers (methylation, gene expression, and protein expression), with disease status serving as the response variable. Initially, the data entities are consolidated into a single object. Subsequently, the learner arguments (such as ```ranger```) and feature selection parameters for each entity are specified. Following model training for both the entity-level models and the meta-learner, predictions can be generated for new datasets.
-
-### Load data
-```R
-data("entities")
-```
-
-### Training
-
-#### Training study
-```R
-train_study <- TrainStudy$new(id = "train_study",
-                              ind_col = "IDS",
-                              target = "disease")
-```
-
-#### Training layers
-```R
-tl_geneexpr <- TrainLayer$new(id = "geneexpr", train_study = train_study)
-tl_proteinexpr <- TrainLayer$new(id = "proteinexpr", train_study = train_study)
-tl_methylation <- TrainLayer$new(id = "methylation", train_study = train_study)
-tl_meta_layer <- TrainMetaLayer$new(id = "meta_layer", train_study = train_study)
-```
-
-#### Training data
-```R
-train_data_geneexpr <- TrainData$new(id = "geneexpr",
-                                     train_layer = tl_geneexpr,
-                                     data_frame = entities$training$geneexpr)
-train_data_proteinexpr <- TrainData$new(id = "proteinexpr",
-                                        train_layer = tl_proteinexpr,
-                                        data_frame = entities$training$proteinexpr)
-train_data_methylation <- TrainData$new(id = "methylation",
-                                        train_layer = tl_methylation,
-                                        data_frame = entities$training$methylation)
+---
+title: "fuseMLR"
+author: Cesaire J. K. Fouodo
+output: 
+  md_document:
+    variant: gfm
+    preserve_yaml: true
+---
 
-# Upset plot of the study
-train_study$upset(order.by = "freq")
-```
-
-#### Variable selection
-```R
-same_param_varsel <- ParamVarSel$new(id = "ParamVarSel",
-                               param_list = list(num.trees = 1000, mtry = 3))
-                               
-varsel_geneexpr <- VarSel$new(id = "varsel_geneexpr",
-                                  package = "Boruta",
-                                  varsel_fct = "Boruta",
-                                  param = same_param_varsel,
-                                  train_layer = tl_geneexpr)
-                                  
-varsel_proteinexpr <- VarSel$new(id = "varsel_geneexpr",
-                                  package = "Boruta",
-                                  varsel_fct = "Boruta",
-                                  param = same_param_varsel,
-                                  train_layer = tl_proteinexpr)
-                                  
-varsel_methylation <- VarSel$new(id = "varsel_geneexpr",
-                                  package = "Boruta",
-                                  varsel_fct = "Boruta",
-                                  param = same_param_varsel,
-                                  train_layer = tl_methylation)
-                                  
-# Perform variable selection on the entire study
-var_sel_res <- train_study$varSelection()
-```
-
-#### Learner parameters. Same parameter values at each layer.
-```R
-same_param <- ParamLrner$new(id = "ParamRanger",
-                               param_list = list(probability = TRUE,
-                                                 mtry = 1),
-                               hyperparam_list = list(num.trees = 1000))
-```
-
-#### Learner
-
-```R
-lrner_geneexpr <- Lrner$new(id = "ranger",
-                            package = "ranger",
-                            lrn_fct = "ranger",
-                            param = same_param,
-                            train_layer = tl_geneexpr)
-lrner_proteinexpr <- Lrner$new(id = "ranger",
-                               package = "ranger",
-                               lrn_fct = "ranger",
-                               param = same_param,
-                               train_layer = tl_proteinexpr)
-lrner_methylation <- Lrner$new(id = "ranger",
-                               package = "ranger",
-                               lrn_fct = "ranger",
-                               param = same_param,
-                               train_layer = tl_methylation)
-lrner_meta <- Lrner$new(id = "weighted",
-                        lrn_fct = "weightedMeanLearner",
-                        param = ParamLrner$new(id = "ParamWeighted",
-                                                 param_list = list(),
-                                                 hyperparam_list = list()),
-                        train_layer = tl_meta_layer)
-
-```
-
-#### Train the all study using corss-validation.
+<!-- badges: start -->
 
-```R
-trained_study <- train_study$train(resampling_method = "caret::createFolds",
-                                   resampling_arg = list(y=train_study$getTargetValues()$disease,
-                                                         k = 2),
-                                                         use_var_sel = TRUE)
-```
+[![R-CMD-check](https://github.com/imbs-hl/fuseMLR/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/imbs-hl/fuseMLR/actions/workflows/R-CMD-check.yaml)
+[![Lifecycle:
+experimental](https://img.shields.io/badge/lifecycle-experimental-orange.svg)](https://lifecycle.r-lib.org/articles/stages.html#experimental)
+[![CRAN
+downloads](http://www.r-pkg.org/badges/version/fuseMLR)](http://cranlogs.r-pkg.org/badges/grand-total/fuseMLR)
+[![Stack
+Overflow](https://img.shields.io/badge/stackoverflow-questions-orange.svg)](https://stackoverflow.com/questions/tagged/fuseMLR)
+<!-- badges: end -->
 
-### Predicting
+## R Markdown
 
-#### Create and predict a new study
+This is an R Markdown document. Markdown is a simple formatting syntax
+for authoring HTML, PDF, and MS Word documents. For more details on
+using R Markdown see <http://rmarkdown.rstudio.com>.
 
-#### Create a new study
+When you click the **Knit** button a document will be generated that
+includes both content as well as the output of any embedded R code
+chunks within the document. You can embed an R code chunk like this:
 
-```R
-new_study <- NewStudy$new(id = "new_study", ind_col = "IDS")
+``` r
+summary(cars)
 ```
 
-```R
-# A meta_layer is not required
-new_geneexpr <- NewLayer$new(id = "geneexpr", new_study = new_study)
-new_proteinexpr <- NewLayer$new(id = "proteinexpr", new_study = new_study)
-new_methylation <- NewLayer$new(id = "methylation", new_study = new_study)
-```
+    ##      speed           dist       
+    ##  Min.   : 4.0   Min.   :  2.00  
+    ##  1st Qu.:12.0   1st Qu.: 26.00  
+    ##  Median :15.0   Median : 36.00  
+    ##  Mean   :15.4   Mean   : 42.98  
+    ##  3rd Qu.:19.0   3rd Qu.: 56.00  
+    ##  Max.   :25.0   Max.   :120.00
 
-#### NewData are mandatory at each layers
+## Including Plots
 
-```R
-new_data_geneexpr <- NewData$new(id = "geneexpr",
-                                 new_layer = new_geneexpr,
-                                 data_frame = entities$testing$geneexpr)
-new_data_proteinexpr <- NewData$new(id = "proteinexpr",
-                                    new_layer = new_proteinexpr,
-                                    data_frame = entities$testing$proteinexpr)
-new_data_methylation <- NewData$new(id = "methylation",
-                                    new_layer = new_methylation,
-                                    data_frame = entities$testing$methylation)
+You can also embed plots, for example:
 
-```
+![](README_files/figure-gfm/pressure-1.png)<!-- -->
 
-#### Predicting the new study
-```R
-tmp_red_study <- study$predict(new_study = new_study)
-```
+Note that the `echo = FALSE` parameter was added to the code chunk to
+prevent printing of the R code that generated the plot.
diff --git a/README_files/figure-gfm/pressure-1.png b/README_files/figure-gfm/pressure-1.png
new file mode 100644
index 0000000000000000000000000000000000000000..7bc4c8f7db547c559fb01ef2a8848679c60e06a8
GIT binary patch
literal 22985
zcmeIabyQXD*EYII2^A282L(Y|N(BRu*dPc<N=hRg64D)_J|L)cNlA*7(j7M4AdN_e
zbk{eR3iy8G{my&F`RjaR{2W6EEcaULj(N{{&1+tB`^n3SpCC9xfI^{8NJ@w*pir0z
zC=|vp9xhyQ2^hM9LSdgYy?ImK{HFL#tNT_qiq`k^4aE&D4Q)*I6~skQD4u7}l(kH#
zl*t5NmgKQ9H2HL-rUswfyBeB*U!nC;@8q6go<%=<$o+ibz}ne?SW_nMA3MpW7X-<K
ziECaDhYas))G(hn%b4@$NX(Z{#P(s?XxS2f)kc9oQ^K8B7E)|pQQyBg9y9MTF=+d9
zkT9^AT_b#u5tIE)Hpf`|NZyz6k01H$YTR2+xVLbyJr83IW5PK5<9$`{+&t~)!S1!9
z^?l=+Y<Y1E-+{;9GI{FgMhsDVr$6Otd0`80f5q?C+ra;L<9-b7$f*wnDAj-Bh!%32
zqk_16b;Q_N-URG@?DIV~LBFPseav>D?>Ya~*I5_n>cwiWWkk!;w^#UVdlrffn1Af^
zk$aJM$(?Knt9UJf)h7J=zr^v%qy2WiIh>|>pNRJjOHA<>=}}r`<&SL$76m`*E4!9c
zP1>cux!zjMmevtvOg!Y{WM9Ky_xY#jPQhiJ7y^3_g_z^xMCVY$Bdi>6UDoN(y_GR#
z^{s!VYI5z%lJ7?GIJps~`%eyIQ@-vN-V~#!!t6i&e=L!xqzQghO^=HZ+124FBKyoJ
z@I*W2xZT%F_B*Rv4yv(Hcq75Q6oh927%!?6EQ);fR4rgmCe|sm^L#PV|5c_hqz)s#
zD*nuvd<ad*oTlK%(~-|=(KU~pXHQ?!B^3>%B{Fr>&`nAsI5TPTYBI&-)kf#=?9lrQ
zn|r7tC9}P+Gn1{CL{j(Z7h2xDpEpsaFF)?#Y?5K$f^TF^-*H*;qGYho1{Gb%=q1YY
zNwmStwTcD-2CaOzPhR~JPG@VdT+-U%j6tsuf8D>_f6`I-^PaIm<ZyoTr`z$@iN7^}
zYc{V7>$cS9bGKCWFT14@U)E(iWyfHA#*D-6Yi~!aP4oVHor&(ZIuqB`kMAAx!1A2w
zQz^PSbEQFrz`fK;<D>YpfUJRF3WxIjdwEHTYFAaZFlp*h?j=rTc*eLIJEQBhLc!*>
z>%lZko<9ah1;01h#@oa{81nk6HOy(FW25tc#x`EutiyT!iuuo+q;~}4rV)4g<3-fJ
zaqzPkJQjHW;k|*B+~nyekMQqZbaA+zLHdkBs|@$cTPuSM_e?K}blX|W%)GAeLQ1#7
zahcsxf6Q}Z$y)w+L@y~sN%z>I%;nMCYN~^Wlt6Ee_h;M1`}{b1ZaQvP1j&TnTl-KB
z`B&bVUF@)#%B?>+5?~seUl2Y0^h53F4?Th$E>`m6W7P>PGo_PdRIRBe3nEO$P;Xti
zc-~t{?(I_aIj0!f#^dtrFyB~LFXSZ}&ctfjHmSenW&ERJ9+fGW>a?cd#Qafj2$QGN
z=R=LE+?nPKnweD%Cridb)Hk;Z?fvat7x!hXjv-WW!amkMqZ8pL{jPd&5Ilx=C5>cc
zP^|DX9tz`MQxrD*gaLogz+c3dU>;n-Oh~{w_!(mu`Oi~2ef=mD8YL+ztmK3-H*hS5
zw6k|((RhmX?aFhdu*A00#8b4NRD+hLQ#6QOZ+^=5O;lZ>)BW^>78CO+tuFz~2&;=J
z-SFwq(koN9t(M~_cC40uGO023gf=o%ZMxCchALKampQLB@ER+yyONM&6QfZWzyHx=
z2xPmp8F&H{kJbzI`=480s8B+TKmSjRrcK4fPPjbA#C>!ZrhN9{f1|u335dNuhrVn$
zdMF;c1o!ASJyta8jG`si&4YIm)1t=C9FBn4HywpxB3)8@{P3AD0Q18S|7QUIS%Lqo
zz<*ZY|H}$2p#ufHFYAB)I+IhynVz2BBjuHJ;F!Ew@W@!?(MHX29AZz&{6waMB)LB2
zMto%m<4!kOT%YZsFR`5nKgD?aV<xdo^pzmvIOiv*D@G#*kN)_{8oYed_FLiXYILj9
zorFf~R6Sq9I1>5_El4fAWBBafOl~Wt$x@aliuvE}e50J=zP-kE|J$Vtn#I-~^2?*O
z^wB!)sWM67$MES7Jnt)l7|*@USLaSG?`lzaSq_$~spp%Vzi{C~TZZbzbnS{#)6I_N
z?X|g(XJ@b1txUBWeh#3_)vduyzms&U;Z7~_p#!Bg$LPjMpcj9BB`aomqB&w`bNN_i
zk1q**vh73@*S+`H?WfLIGpo=YycloP*K6ihO^#-X!q)nHyx~InQ|IOJFmJ45b>+^>
zHrQd(B!~Vwm)u$l)m!dtD=8&aUwUhQAlLf4Nc!D(i9bAM51&U%tbx%z@Nj;htjL78
z)V!N>sy*$+)DN{>1IgMjwumXLQqy*(s4J$y4Y4;=SFQ;V9>gEw7no^s@!=J&>vgYg
z{+q0jay$5<PNfsOky71gDJtJ%cPrPdla-R+>Fx1<E}@%Zcz+yAN*b=(S&}mCd=oZw
zP316*@p3GDq4AVxjuSAP>B=%viKV{b5J~BQd*V!JyO_V^UN#|GImM_k+UU`?A1Sjz
zeK<#ARvfcjTqwb5*2pT_e6diRgZ1<k$FNxKxO;1Y)N?dQyrDtRygNIr)P5$qtmNju
zC!c-M6VI2<K8_Z^>{ZHYXP;{K)e|78NIPDBxW2FO`d#NMjL@q|F9fFrGG$}=a^1Hb
zT2(D@o{fF5FMsU&{cQIM5rwm{Zrb~Nug&E6a)eEyxC})#1g;Wys_p-KoKkN~@ncI+
z=ArQ>V3S|eW|tFK4IQc+tB;7d_3TV-u3`O7A!9E=cYzldiF=~f!;WLWX5c+~s->*3
z*ZxvVB36i<u6^P~n0t}{$?kM#hEKGR-R{<EuH^uqX;-FNq7Hor(VuD4v9y%hn^Rsj
z3oYN<nu$`-$}@WAw%mAqX(}z=NIl+{Mw7y8T6BePOCTeSc1m(dT2#nk_FG{5BfHse
zm$^TMwAb#zu@e=~e0R7fukeo32x`7j&M+z^5?uSHuaJPoAayD<ZU4-FyWFHT@fE{F
zojaJm3sa(FbSiBbY3p^%{<|yj?Mt7}T166mF%>F2JS>zr`8gANyW1%mMV3tAEf*N=
zHS0%fLsDGVe^NTid|QV(M)?Zf%MoLJ^6S8&+wjQP@3z=&-msqzofN9bdwJYlNPzj9
zJq8PDtE((^&L0FcY~&ppO4e>6Uv`qIZ=@8ZBu7lz%G_so+x^cOpbf&-r1Wx;^=iU+
zQ}cSSU2I$aet&Um!?9@Q7a1Aj>2wEA@<PAIFSbd%5X(9hu3;H;*0NZboI~9}wod9|
z1;bcf7<Y}y;>ah%{^AGtj=8Ui4~{?}9%i;VQ^#<gWC+tZzC_CHu;2@dsb42slNw2J
zvdez*nRaK#O251&V_C7P!u?73-;>m*zVqKXc7V@l?_q0pV{Yv%H%s~8o@iKB@Z8^Z
zEZJS_8ykOVb?wEF+e}uGs80(ew~<7eZ0u-3x3*LyshHTIX^E1+0e>S$)0Q6D#5+ON
z+Nuu*!ab3wnxh@nB#E4-4^K`+buIn{5hrlK7F1r#3)=-dH;*{yyvVLAp3K_19mZ;?
z`N{QQU?^`MVrwlLd1|+g5?kYTBBgCF&bx2XKg;F?o!5Gd8%49{`idf&h4%Yb2;Ju^
z)<?V|=?+Z?&wvaWmirB%tx0-P-E;g-EddTRJk|-$9@`b`gKXi3Q5wrzgSM`M`j4Go
z{BhBEw1Ih;U<hGS+pb${n>)ikL<XZ@gL2(AZAup26T;H6^0G<&Ji@*hIIy+2nSK@U
zk)^f*^X|-RdFKw?0GjqBw&r!b3nv>`Fb_7%3;hqk<9ovoSdOkt!j)GqA7CF{`Ojzm
zvp&C_(Epvbkxhi=s1F_}bF8m)-xkRc5I;JWA2qAPFeVEv`V1ypUMUoqDc43^j1Rq_
zc<XjJ`vrhp3OU*Wlw1bBRD5<U<B<lwhbMEGJ@AWj#73|F@_1ugq8Op1q-2nGg^MJB
zsT5JaQ{#Ds4=wwP@1)Abb8;Is1mX~qOFTPsbtD5b6!+jjO4uC09FF({n8WW*e=8uc
zDfUMFBYaYWZ*T8mtlwiiI*Cnq#Xre+>cZGkJooq1aO3MDuUlzTf*WagdHtfXQtj!3
zXQOGQkcT(EydIICpU=)V+Vl1g&Y7S;bnqz52Xw=w8Vdkd$WMqk&h;|7Z>?N*=eI{G
zWNVf*NNmWwyjGhm8Dj9Wza+6}O6|bMVqlL*Roi2jfMYcTY{=!bpi6m4*DHqC_Oo0*
z0L7)T2%Xy*Dp@;)?=X)BWQ=%OR@KJntBrW;q|CBk!#OknBFM$@UpURG6bQ~N>Oz!F
zcv$~mVC|H(A!<53074Z2lbv{gwAb1%I~OiKJ=_Q4c+BjxkCFqVMQP$GNK-@(67Po)
z1_&WE9<x>N`HdrBB)&SxqO`{O?kzV9MNGtd4lV{z^V*0OS`Gayrnzxs76u4Pd_&qg
zGr67__b|D0`4Am+yuEjmQqaxGC<PVIXFtvG$)AGg?PKL5A8bqtw@yd-dbA`68)rlq
zIMQtavU$2^t8Uk#So07k)<OTq;f8v>@S4GtXG&b>^>GtD7`i6Ya49O(Z&J(Q@g2>B
z=u<5YEB-n>-fz;qiH_!Qi74=kxlpSc0A$VtObco4LD<YYDZ51Vh3)Vi=q3o`bX)Nr
zTawx3OSd<-=c}Ie&-Ub{n03A}T9ALI_bC|wc&^jpNQ@$*@xl90-t5SJ?QQ&c(Uxzj
z$m_n9pWYG7C^Z^wo_)F5vzu2${Mi}C$AlE&&#7;mdSv9>+j9t?M`OHZENft6d-6S+
zqOaD!2EZ>1M$3FAa=dnVKf`t+*6}tZ0){&qi#D5M5fd?-yobODZv($L8^>;{EhRCe
z+BF>O*vYS*sycHg1O;a7OPONx;|_5xnzj-fhN$hf_=(G+_X@AWY-G^=fzkpK^VbvM
z<=Oe{QY%5x5CorJw7Bi_$6y{@2#ky4Bof+rTT=h#*<Bj91!2u_?mGlyr6@b_088~3
ztG=d0XtOG&N|U&ADP*Y^Br9jAEH&Mj|0+)7&M;7FPf8tW2hp9W&N=bipIs+r^agY;
z0hnm)YY@H7_FU1+8(9eIc(M0fAzU&(b7sp+C0#+Ra(7L`c|6KE*=ndVN?owMt*rsy
zq9wuG^mJ$I0{ejpp;a)&Uz`gA9RIalPKe`$&|a})k)ZpQgqhhJ(}mKx!oKnuXHrVU
zlGW$H1i$`E<H4f;<#`yAR)HrQ9j@^i$84+8<t!e4_;oUn;EyFR<a^1^-ObUikgiEn
z;c@)g&t*F<4W?X7>~(AJ*Fn3-qL!N+<&pm2$LBcZBls^&INMvmR{w;pV&+ZJgc{w&
z;ZnhbAZA*zKei&F;(g|-nQzluQj<+UL&@p|X5QD{aY|!gwp_yCrTXIm`q|{GJlxZj
z(t{_>KaZ^sc}5_J@|@GKcR+?S&3O#fp@%O5F>3#uL3A*IiB9vy$_)_$a(0#!sfci4
zAKcnF0oMlU8+>y~g5RDnE@9?--5Lxf`*Rkm#V~OQ$O0j0jbdg2pxTlDgj1(d(`9XT
zDp9>S<8{7Q>5SVQ^*>|wedtyA$YHLxbH`s~iD8N!P57D!@90=zC$a(Sr=_DHesCy>
z7f5JPs~F$ukIt1H4jwvbMr!(C<4~A_0G$MBk4`eNZ#D|U#Liuk`p^wli+iCujy-yD
zbW`hTTGX@bi|K!z06b!^Ew2TV!<%SJ;lZt|jP{486N3f=Q=W!c@GwWi6IJy>T`8?h
zetzg0hw0I%X{`6+hc|go0@rz~JEG&zePFv`<DuQr8%J@CSPQmoax0kcaGqW-P#FL7
z;^19C3Ak;%%~UVAeUh4w;mZq}b;+lH%@U7?5c{-OYFe5UK=2gn@9&Yws#fNZb)1?{
z>1Wlx2d~YjuPx`S_u=rpsy%V2uv(Bvp{%OK2w89l_$kR(B99whMd)~xltWHn(=4#=
z<#%)ox&_lYW-Ug@v+{mlVIy_|MDV525V`1^E-RDg#r(;`FtBia4b_5!gGWNi`i_=?
znD`5M%j!(G6dZ6dAGx6vNM3UFzntvJHw~F*5<J=(T4MR?%cv}stmx#Fl-i#IWxJ;$
zxeT8|<|fXnk~vxQ?DWyb;$fe%F2@-vw9r7T9z2>Dk}L4qjt7`zOo9U*dv<*=aInHv
zs?u!}Go_L5aGx-+cd)6j<f%9JYMJA|R(oT~yiLEbGS}D5(2=Eaol?LhUnep1kkJ5+
zn+@X&Y6U)*Yt(qg_N)hBkknO04gLZhleyl4yxS40e-Dbx`Z#)`AxdAJzsPccm2XmF
z|0u(Sj6)0Ki!hS4ahcNaJ6$Sswx&9?1Qez=2CUQ8>dec^cn*!iQ1k4PTV@@v&u=bI
ztTQE8*;$`DWI`Yq)Zp>pCa|ew21aokN9Qxi#Zg_7!EEtj&peED#CWvB|9Ck}cSxS3
z5u_THS2=$eLuiRhz4U2XT0%zg{#qAcZ|-Ag<r?_$ue97$OVhR5gru;kH1^qF6Lk8u
zcGgOjZqChn<EuGUd%^oV6GFk@_t*Pv<x(x&{#v@>`<Tx$Od!!@Xc?xj&m9Fv9||*z
z;czefYb=41w0-VqMFoY?z3ur3)ArO_V0X9}&(ss0!J8kf&{&)6qo}`MOL%l(Ut;#0
zL|?7aDs!j>?ufA~q{FJOQ0B==YE}p*r`_2Rek9qTHaIvsniaMy<owMiqbBk(jkY!S
z4yJ-uz~dQwI0P2h5Pg-6^a*Zk@5n}%hLt1|l=MD5jVcsh{xIJ7G6n29k>hQJ%W5z%
zB2s>*7#oZ^+CM)#Yq+&C#ijrGxY}D??<gLj=&|z_hfC?*0;|*l!LbgDh<`X@ns*|G
zek(M&HA#Hi!hif}5WRSGcr$G4wm?vA$oyZ_8|Ah8k;Yg5hLE(wPK1?G5u7$V8xN{H
zdwYukXqHSS`s=Ab3u9Bgy}!4ccU-XG(qD|6p8B=F$|=_5V2}cmle#p`O9090(&9Z>
zZ#c{@WIpXN2XM{E!O?hG=)`mVo~zaG(0_4sy4X}!H8a{d6>HG})HhT-c2`w5f4mS%
zhUjH8T#Zp1d!xt=^76zNGXb4#WsX*@v`WX&t5fY?Ru(%jBuZ??xONo^ny3Fy`1UeN
ztH?68-+g_+=Et+^PYWNK(KPToFMWl`brH}(5E75NvNZHn*ybEd|2I-E!V%ET7+{G|
zl`Qp53w@l!BZO`ufI}g)w~=IO=myV~mNW(i7gCc5;PRqs(E6VPf?{tt-iG1nsf0vd
zHIF9w$t)XliS)a9#zbuhbBf4g0)=!1Ms7yM27;Tn4r~R5c^)603*=E)TZ+_6(SorD
zFan!Tr&S}H`FJEXZW5kmB~wTZ>Q$RSW!*`761A^bhBav?eF!>Gn2Q|4Jb04K4mzcF
z>FnzH651|Pj2G^{O#oOVyyCpjEV!Mi=`xj~kZ;l&*%j<Jsb42!bntq#C=Qhdx-Z|9
zURB#UCEs$O*C5j;=qcF+y6gScK9h4BuqR^qXj#9@Bo!gVhkn<41KIk54I1~Apvgxo
zTd$?&^waC-iiS8ZR_)j{Xb2%JU%F!2D3C{~&yBa`jv3&bLD?7TfBVgeA%35B#EDQl
zXysjm{dC77Zag26iHl1qm7p8X_!GeS+Tv?7JFjH-^MCf9segDFCy1lb@))n1u-;(2
z=Hw%q#G%ln1;-x(5cSKB=}ow=5)fLdXU5tCfxfEuj|rs;%sNHdjjRrj55|LX92%=-
zASwk^ISw{=*b?0EmBDcgLpm2n9qZ6T<^?AC<H0$A`4jKOe&Gq)cQ}BWjpE#jt%d~C
zl{3PfH)Qm>c}mPC{_&dTPe9;l)UlNT$DwEb&S-%&3FgFZU8Vi0L@b{bkxt_ed4E8X
z*A1Mm9G#1RJww4CNNz5TjUjmru*|g^i$54uvlE`=6z7ddDw+Krmq1_1CgAz8LD%^(
z0oSz{i|YqY4jn@<`|aJ$CZT;dB)s(7cWb4F@`>pUmOCRX6PMe@0uKl8Cjv1Dew`mj
zOwYAY=(yck>6!NqAM)64%zrre;0v{pEih}1k2by645|OD$;WsvLBj{tUKoSJic+s%
z>;jC<k8vDyHX@G$%J-8_&4oJ>n8G*-ux+pbGS+F$J^)ZlHJ!dc4i@s<3*_?u*aXxz
zX>3Lw<R?biO#!Coc&yQ~ar~yGgp*q_EIOZ;77o@~gtTRT6ClDSb#l>QUxB&tIU-rq
z^p_u{r;#^I?Fg**(?c$P>2ycXB8|t!t4;q=;DW>9M19H}31N~6_P<zp_g26^=XbZ)
zBfud$ZTm>a@U-?n+=qNtPvsofpXAc}EB@q<%JumnnoGhKCT(-<KPjwdYh6b^;+IUn
z$q=2u>a3FBuNjA%IIXI73-P7(YZE*%rV?ITRc_uIbe^DY6@0@V#1d}oRsK7p_KiaW
z3p!eJjAo@?7oq`@&HSkc6wASi$Nt7MGQgSNe<(Be?<=Vn=L8xaFZ3A4`w3DAbopPa
zf+|D5-H8_RR&Y;3U1GFCe@xZ)Z8dl><DYNM)`fQChDvPHDE)0(#VBes)$(>qhs1bX
zgZ+cNesDO&5(sa5rlzKDolXbrlbF<LCoeC{AFiOGzk2x<gt4nW&j5)}s@^&%QFysx
z+e|cBU^YWuq=b8tiiMA+L=e*rsDd`2N7t1e1;;5P)>6&I!6D-?Kga<fz9}^FByJmd
z>(kS0M(Q*(L}$Frs+!B@EBivZyEa2;{uqs@ELuazKJ(u_eMxCJQTY~qg^0_8jL_dC
zZv()DsD>EgHLc1QpG@ZU*o$n$XxD{apbue|Ynsvv6goJMUS2+4`Y82c>wBAk0j0T%
zabK1qD5~6HrmIn!%v^Bc;El+65=PJp-&hU!+y)CLYVXvn_}-E9T@fgjP>Yh!9yt%M
z>V3r5+Kx9|=7<CK{xr2+^8Vc;ZiM)2w%3m<(*Glb^1k&Kg!;bjD@7gi@vpa_F99-{
zgq1QFZ-`>b%Dfdt>V<o=QSiX$qA*#Huu}l};iOTRBqY|eSkadwetdc1U+6pS#qhUc
zBA1?63FvX*Hor-G*_J0K&eoMo#If-$Zoj^hlpDhQ<%k1ecn5Z|0q{FAJ%WIYh4$u?
z<3Rv)`~~S|fI^5!i*vaRzcPRLmq<=e<jqmqLHra}z0_$@ucyE~VrTISjlwGt9IIYH
zq9Jr*Prv3h3$lTq2+=HgXePaSuKDbtQ}rGtvko}Zouy%{F2Ic&f9}mpcA4xNR5GWg
zrvA=H@o248z%G;~_B8<s-v9Qt=He^IA$Qd`Dp_V*kG3ofj>0Vp_$NRK5iEB^pcZ*A
zZa2H;wQPN|PmssXqIeX~gG9UeK`m{k?>JU%-8!BXp`2R1+fmn#mKdS!kE2zy)ENmW
zI75L{j|!K!II+skcgYkCeLUV+t2<i@A1(0DQML^H1z3lw=7YxzR8B(I4~`4O9s|Yd
zcSI)F+Q%*(P8kCm6H+o<gTFbxd`QWX7^O;&Mu7o|CRXm-Q%CSp9mxlS8MO`{<n0Al
zFx2iI<wC?9kjR<T26G--1{w`l{tNu}4M$;^<Qqs*96dP!uAs^Ps;!DjL0+kzo9urS
zJ>}sFEg{=c`jp5B*)oS(+wbUxgg87FxDxm{>}V3Lry$J@XP-LCSh2}yQ768>zkM_b
z6MV=|rNZrwp8Ve}&ypzHy@OYIe<b;qYv|9eRjPy#1wdGajePQFmK}R7*??n$@r;+3
z^7jjeIlaq%mP(zucAtfmNjfk>+ePwrD9d>r_tmg6R$<d0x4p<)iutI?ezUXQ8>fna
zXc=AY)`<WEmx82I`g=ZOKDb6M<0dL_MN)33Bio8ayvXHMW0(kjFFSe`XKjAS>R~f*
zpw2#W9AZB{KQrvjH|5g%==+W0DHW**EZlin`mo;#2<;MP)$ZC$1XW>dyE^>rSM%80
zocL;d`>Eu4G%A`bE$a5~u@je~gXqLWHH5J^dRf7aThTA?{_B+lX*oT~AJQ@~1{1Jz
zWemTwg0}12RUt1Z6ia-0PQBZehv}`0{m&z^fK%JQ&vxJ=OazMqpp%~N%3?!`!8hj1
z6_KnR;hJU#%2WqkX0jNRUeyH=Te}m!HdM=XUAJu2z&JpQ#A@EaJ4pc0+JObK84FXb
zles>?03eCW+a!)ZUtdh+hO>h0SXiD#-!&-I1xUy8DX$<%u7R?>k_9rLFz_7pZ|{9b
z1|p!=mMQ(hZzH{P>NQm2qp+^rt_`_=ilCSiHBj!H2NM=iP*6yQ3Z6oRi#_-GY>zIL
zmj6aXa4JsaO~8Sfe#`-CwFOCJRGJ`>k=onY#HyPvH{cU^B=}BPL*#EZ8@N?>v}(5I
z1&I93TpLR2b5{hchtW|$uJ1@Fo;oy840>Q@^I`RZ!^6j&w#Yd(S$d5h|5rlNk+9Nn
zelSsZGFz)GSuS2M^@VY5Fe895%~YbvRGF6sZOM{cjz2Y9`&JIt6Qa)|Bl6_`MjycZ
z=D*hm;EiS?)bZFGwY*fy?;Zd?df*(GWTXkEgR%a~5n9Rb9C?244KjpFf7Jl>jS`%A
zNIU-@$aKEVot9TO8||deJPN;{_&MLjj8^?mA;ekardzNg>%o%Hmf%3OwoY`V*1bmI
zpj2vIg)F^^oM_d(54a$j@cR+fDVPQj$EN=U^--YYB7_a@ggCU~ssaIN+L<Yl=7z)l
zuW}-{<>VrOCoxbVSeOE#hz%4yQ5tt2A8akH^)-b3jPsj3cH)d(p^KygbD~vGZpx%k
znjxqFQs>j#J?<YdcVF3m^Xc62Xu-pz-V&kdVceEwA6YIQxX3^?gk&cNlASwV^smqU
zVn;E$YHlGU`_LGqBSy+V<-1>Bq&)JR$8a=xNp_)ZgAkyIGyoI{QNR*yl9=h-q2_2l
zitWTYNL9CLYnqMpF2V}~35wBE;aWsluvf>H^D%-wMfi|*#A87!G=LmaH6MIB&1E&n
zulVZwhkxrs<ODAOkWM4Jpzuhg*()vT(AA=M@Q`&Ce_&Kz5EmRwCnvO<KlyQi@4CsS
zr>7Ae&mz6F<X^jh4<#kH#{{Pv7?z{V?s+;v&OBB+lN|zNYaM7E+^#SGHAz{7ai1Pe
z?8!5J+0}HKO~pXZZGFHo#eSx%?Ts=Ukoh`X7C*3lFQ^}_^>_oN^N1hwX!&we2$&_p
zU5J*Yy`vpwLIgHdYrW=%0Q^%wDQ>%6Gij-wiDi~M7X|90ew%s@E{`22({FEIIi9up
zJzNQFj5(v!XYBqI7q4-@X9sq7<#Oi@$Ddd9=}x_R-f?~rsaQ;3a0bxPh_?ZigN@i%
zpdG3Q@rv=o767rE7RG<Df#vB)i2iQ7tKE8P;Y9>|)Zc;8X^8m-9|{BN#I<M<*W2kd
zoR~L}e8)^68_%_?SZR~W>;H`V+w}Wc-hy<=8Myx4ryd|QkN}=TERB@`wff_p7!^Tr
zHjxR{83l&lAJ(I@9{ozvs>|JyxGO9mhjBbV)*kO;2!vYx*b{ZX4Wn8OJNlXhD~;#Q
zgK7aJ3+dk_l=x7`13~iHhY*pV-%EhUVBIcrz1BG1oryy|f5B4Fc_~>bTeF8h0N^#}
z;A*)rZ5*J5ADgrC*M7g73y+LLU1{y-fVq*+1?}=;d{OwIg}Pvo)ez^FH90wUKkaa@
zu!LoGO2F@6EvlZNTk{ZMKPBblC{FM}A<qnV?3CLr_2`wv3lP?2wAya{PGbo$5s?|I
z@|#G{I}&jWB)+AnqTwW2Y22f*PYI*pHapDGfUKMbq;iLu3G2Xxpu7X%3guO&BiRG7
zJs_3Wq#+mw2h$6E3$6s5DL*>Z@GcCK4!_#NlY*v&h-h-_#NZKIE2;$nclx8j5vv$@
z2d*UH&L1I+L{6BJy&l&QDfpG8I&b)J%-+N}&#8FnkQ2zRU4GYQtd0>9lCYOnm1;|)
z-`TE--gc0q!%#<uOd2REBXjcpN)RLHu8!2Z5YPxE=Q<Fnu)Ppii%d>UtpkdU{v~Dl
z`9L(x8sCm9{a1YkJ&Vm=AIz3rEV4A-$d+f^91h&F6y(ANLzV6dMV4Bi8MB@xxG9TQ
zEbX<8#d+f%*)PyWD~JtD0+Xq!X|CtKM-^scypRqM(JZ|p(Kj3|tL8G%v^{`>WY*Zc
zwtlUPft2q0&jE$P`)U{D6TF5NfjqoALqx?h{END#6+zW0|6VQhHS8-J{bi0hz*K*k
zpM*+sJqXQ`@4nOX=(`InJ$8a7*6(dLkfQb6+ewW#h8%hnnDD_XE8u13#wxde{#Ywh
zeS96yfINZ7#E9sRM<vd$hB6r<!vS%eAW~h2%JD6urWjMq7Up<2u1ULZB;E$3v?!TE
zZrR_mB|4@+@MLf{-tDZk%jKH1(s2__22gS*0}Dq|@AHUQLKN&nls}^bA92Va`i{`F
z@tYKa_=hL!B`i_H`;OwsmL37ZrpK1#I<nFl0G_xLAGMFJkRWiCnBjLl>f;M9!4wlj
zW0wOlo7fX3zvw8-LV`RJeGO-&HKM5B#y^x@O+mfzJ|Y`f>QI#P6D*1aXfPt~b!TDG
zmXihl*QX;f`A)e5F3bgJv`m%=0z65;c95S61{v0eOUCiL%{`WE_>Lzk3V_u2QZU}H
zyJQI)qPWGwFX~vHdoksVZS?*9W0{#W!I$pS9qGJUqD$f^?|n!lY%ctQcnpuK0R)e2
zPQ+8hQVlvzb?3xKu?#v2#FL<DjbZM#pJO$D9RND^1n;ypj1&2j^jugtslSOEQ5`rU
zTc1aNnE^4DH;|n55QMQ#@|3Q2X{3OMyv%~51{1#o%TUAej_B72`w2lecjt{#Ivt`m
z1qw5I>E}?Dj-5NgRs(W-pmU7C7o7{0EvEXY)Hr0}9uiJ+I8e_4$7GuhC<akoGzqM0
zB1+L6D8%+Zlv`MqTkU78b|d@ArxQvJJ8K<V4b=xjn7+-@sAS1(ZarcFcd|Z2Ue+Jg
z+24ww0%4W(r%(tAugAqlV6RG19EWNqbRZ_wi(;-=^eXM`&U@DP7MQ2p`|zX<^bcOU
zSA3+oUJuw+%Ebo64Rf<_I*8DYLxu5kOw4guEALU_e}p{B`Io-Z8)-LztTyuY3+fwP
z(zrr;AMu=5TE#=K9;(xMdBwx)JQ#i#XYf?fX##Du%aAQo$StIsn)xRWp7H$tn4-bF
zc+;966R%hiYJss6&lHf8`G6y#S$hcJZprRX4S9FgI{>E*Kw?|@asx1}BpqY~49wAH
zNJhY#;QbvkcLP|tT)j^Oybnf1m2?ru1eP~A706{Yf-ehBKK8*?@N=|&u;YK6lY}J(
z_2a=LaiS84Wri!0tw!1nNPq-iLw_<!kxt8fwbQ0kS#>kT9YFn4!K8}&J-NS+d+qsZ
z+T>B5<_^e0gEi3_)T@Q-fI}J+Xf0lKAV-~h*$}Dj0mh(zvjQ+i0w4v0PNmxtFaT2C
zm^h<8M1p}3F6xFn_Z#t6ea%At!Tbq_%)WawF(2wqx|p83s}acYwwY`>&n=WHSj&8-
zzG%pUt;}JLtEit_kW_1;z9alj?2YC4{T*Y`ysoZJq5YlqhT6?u^PJknA72_77G2k7
zFLNZ2GzNSmMpr9?foAh*k%8r$lGJTY6dS*04`q27&swB#1ybRQeC~HX(_TJhuU+X@
z(hn3SMtf)9XwXSP0mH64^F;g$UYiqYzoZmsRNYnNK<udMLLoAd!+z`AE$Z|EV7lOR
z=_t7Y&(D~pQ525od_Y4jbFNq9by0eK#Knss2y_!&?i7sR+r5(@rg6`o?7}G%(itRB
zyvDnPcY!tY6{8DCPFuV1El%`6ox<dpvF$ua3G1&9x#`DACl(DQR!|AJC?QhqA@_By
zaGi%i>^@Dx8(>%UuzLuAvH?FJ!9;T@3z16zlVPD*;Hi;MakaDW4b&<w7g<9hLfy%J
zbr(jX8RxbnNlFFe6Z`b000=x1vwWaX6AlSr<kl^=c0_1A@g!pgczDZ)au(01F6p9`
zay#&Q#h+76oe;cAT49*ASp}{j(VdIF#3b+rl$cs+`y8iw^3uQuv^rh|KZEj~#Q#R!
zu2u2P`-$QD<=ok7j?~q+8l;&}33wcvw(G=KZ#nmtvvNZ(1KeJ?dukSB=WMffdAF-i
ze#`{orwO81q+wCz-L(X)kO>H)xqJ+>t9eF^xwGOUY9OV`ot;3x+?gdQzL)k=8FP)6
zFav^)3Y68@Z<xG$N^#LL*0zba3-Z5TH>%S#rU5yi0$@g(8h#>oC{<(BJ;vMxPSUC;
zDJsiz=Lb!=Kx<KL`$=uMBXiYdAm6;Z$*e8e&+u46*Y5g|)l&O<>)n-!UJrvL<vb%9
zDbHK^mGhO`*|qKG8cr>;ulBaYS3vQKPqA0ss*JHC_gK4uI*lJu7^`ybtnMXU1&8{K
ztHw<+UVZP?Oow`}uOcdY!}VhY;hb7#Jv9~f0G|I1VgEwsg^s~Dk$wjUnukq!d}K1l
zI$L9<eiG>6{Mn|h@ZEBqWkTgD{uO_b*10>e;RBebJ6*)+Pz4&3(mkL6pXJEFeW*^6
zb@K}O3H4bO^7^>#=9fE=pc^dZYj~B<(5-$xr+KeUR!VA_$y83zqN~S;=GRrr#;d00
zy5SGr#fw>BDr~Fhf9@o4-iH!&BtPhwCuN#F#tYPpUu>*q`*)mmMg1$e?Z+JVH8dG_
z132rvM3P6nS%FN|O_8tMZ*+(L0A`@Mn2&CLE^Rb2-<>qi>Y2t?UJ*pSA(?ybh4Z-o
zl%f6HRy$S4UwlBril~-3Sj+&0aWYaAq9aF{{f*&5Yt^PWfxeydCUV{e9y)@n9g5)t
zrPP+WRDcv8FaT_O9P+BhzmRHDXYNe0Y5hkK!kCy6uTN*>M2AN~>JmMygjH_B5rl^>
zbDX#AX%<;yOT?AzjV7P+YN_0{sO%`PuXQXL3)jemJHI*Fe~T~pGz&YuE;Rv6$fc~r
zAj_~0b7758ERJdFsc`laP6b!hG=kkIR=dc{B|J;a*<uA=Bw9Iu*l))JVgN>wp<~K`
z)(6e&l=~)&$%aBktuTu43H^`S1!H^}<kkDTL;D3xGSOu3%CDTaASs9R2P?J2W7nm<
zm&#iIAGZ0DiF2fQTTHL2>Q`RKyDF9c(H;on7oSJj5708qaT^|sOtp$%iy+9KT>jU7
zq56+Z(A%NbCU)%uQc}$A|JjqUd(GY@$T}Pl(;aP(O~Wb%yGvfOWLdi+Wp+~|MFM4P
zc5f)Qr7jM$+@*HUXf}F2n&^%@r-4H1q*VqPsyUWSuev;<DYWc@hSp}Y-&%Fo1)jf&
zc`iFd$Gl2)ZipgXEUn;z((Ax)SuZ>!dk0Q5TusE{pQY`za&=J6)UM#G_weI;Eq1JB
zWxA8Ctah8)xM=paKRE}3)nK_t-<c~Wy61`}i_SNk=3~60VzgyPJ#elz1u8;`-LeT{
z+db1eQ9;^qu0MG%sAk7R5h{9KTXmiS0b4rKhk%G2-jzKQ<&$bddh$kYAvrtl6+RZ)
zjA2E?Revb{dUGLCeykSTW`S;L{}H<<D^gRCf!g-YW%jKSGG)}T@+F92y_DcIu9vH7
zp^uFZ*J^KIjcloAZ*%VDXiuENRkw3b7T+@4wAZtuUKx%0))!KA@@JbT-y7awe_UNP
zfs$KTdKFLTOD07}9Q^I88LskllJo^>UwK<#b^0YQJ$I_b{*)E&WIuEqP^$jBJILKq
z`|~~Rn#BVvm#t%MApBrivv8U(m8fW)5jjRJ2x<nK4=1iWO<;7^Xh9{@Pw84RB3b*G
zhL*P|d5|9>5FZ0>=s9C4^fGXiDK|f;n*EvJPO!SB&*MLE%Ge&1Oc(LzY>jj{!l^H9
zS`ZpP7C6(Ni}lv4Xzi+yE8Vnp@cIoEtJS_`LUk9{AxGz_lnD0iDVFfT_s>T``lnZa
zGID1NH8ZE5l{E04tai3x&VJ5Fw<v!*WxIj9IZ`7ts1hf~U3N~$gT3supkmz1>sHt7
z^ED*6vxwGx;^?W9^@<F(8B<c%B<!nJzh^J3yeLegTVhJVdT7@YqpjF_H-A2}D@`d~
zA@xRGg0PQt)e^`e#OZIeAzF|`rr{?V3n1gn2Q-rb$)^e^`!gZih^Act0Faj-AMP&I
zw=&jtk%c-Y8H+HnK$H&EiBhaunC-I;U-N@<ey@%~caI{uSjwcsI%Y`aNzu0&Y8)q2
zX8U%+2SyBVrq7<SJP!=3=~6pMheqN3%iQFX!}Ks$7jnxbj2GRy^6f=sivP4jS+FLT
z`9t9oBARp7`_`(KP+S=8F0#^5WEP0%t&&>~=+fnx+t0MmPZP^s*;-#Y+Y#?)!X8)i
zfR9USJ9650g^;tjx`B)M+0GRi>hsz5nHog^HIR(EryWP_jPEV;*8I6IqUvjN59;HH
zI(D@ERJ4Cl>%VCB+qq}%_uBTVt#QaL<D}>)Q0`Bm13+VW>%#he{Cx&Yc^OHK@mUQw
zBl)nC{B15`+}2(Eb0sJaYD4p(x~o2Hvl?D^b`^@OYF?&4%wN5&5wSw_DD&37*(Tk)
z$*=c!F*>IFW(8Bbpw9V^pFy7VP4QH7)EjQ@Nm`!m@HhCydHl4PXJ<?38dpc~u~%O_
zHR0r|2pc%oH{F`_uVM2GN^5Ji<mGM^$z3dwV+>+5MBx$<Q7Z_Li6OVXiHc5?a99|x
z`*d>fqop#bt3mmcd769YEXk5yGcV=p))n)v2L7SDI)}Wgw-ZUr-Yg%nhRzTB2LrUp
z0f7!QBJa{lF7I<?PY57M1sniK6Lts|SeiBju7@6W*h)x1viFETONjtU_7DO9guR$8
z|MO>5iA3royPz&s4;YDKt0P!~eJh#Loh;$`sBtZOLf6)}c1YWe2dkM|zr^;Ew)hM7
zhJMmPduDeqF*1Jg)DO;Sow?;XgzfUtEV%tEXe}F(?#%I~g~Yzk*^B!^B)*ky<3U+s
z{aCQ4A<j~svt%@w>SBd}qwVvEnsgL)f~<(z{UYY{aDb}*0^i-qT^|*y#}8G2c!bw!
zLD%FMQ_bKSsV9SgfWS!ezAQH13G|%bq?nH#xhIBuYH2Ts0{gzn-6C}#&1qtKAFNg#
zXRCtLJ-0UW!W%HR%I?xQYi8bk*Kl@&;Y<t&7_^>}b433*1EsMdq@d2L{XFk<b?35I
zT$Y|?AFzD!wLoH+#(B;a=8Bh<v#Dm2ZMVMB6T)HZRV(3i-}UbU{lu)_f|lyqU9!6x
zZ=tktajV}RUbXf^^OHv~dHXj*qs#HL#~%9b`l^8_O2#XzwxZ4h?=d17fix|mpBTez
ze=_Msw2=Ba=TV?=)yCM+H1awme+lYDfo{0+rkvrZemD%$%W1ave<_h)BeKP?h>yQ=
z)?XX$2CS017?K$MF%x2CFo4%f!8`{Syu5OtK9YlnV9`HTl8AJ2D{G%R;*b_j>(aUC
z1W|46K-t`1mQ@>aE-F^x9it36k=|1KjLDBZ>`^%Uaw2a)RyuydjSCY(V>q<VEiojc
zx_tzNGl1&l$G_m!2<ccd<VwiT&-OUGzF03YFnChAp9&VVmx)jOG{O#Zy8rP&FbtMW
zLN-k9hJ9J`(1g*|YLE@~_zL{CXk^rMND)m~I`~l3PHasM3DY$_qCaL2M-LBK8{&}U
zqq|~YZDOB&!}>F8U_%~49z^ny;lklv7{;I;HKt!XTpD8E1QdqJ|Eu9NScl}Yv-L9s
zk8c>{@`0{Wt=aq%dCjbCm-L8TQe9!pG75E?3HdI75SdG#WJO<c8#nQ&2t>LZ%X^`J
z>p|hoOV?`1cu6cCrYk0|;eY0dTf<KstO>oYFohkWT0Rec<byFNv2SwOP0I7mmoFzn
zHw9$qj!0^BGJ@4Om*$E|V8<Kf`uQQqKPzxt_IKA2`dQ(%{CPNVQlrqdw`M8z@jBf5
zJd{<rF2}4!^LowfkBGPaDh7s0-BM7A-p!`3sN|QT>ouu>?nFAI{{XaI5&+dl;ar5l
zZacrR(DvD+BruAX_A4F!0JPTFd%}yMkZ(yr*V`CmoM9^~c2>1MC6;y*qWMrNiUX`8
zg=n5&Fkv7=an|CuQpvn~9_guY=*&=!fZ@7cbJhvzWQv^6*4-%&tX=y2{Ii;k{4l5v
z^Ub<4O*!tqtyV90dZ!FZ=?v&53Ex$9IT;RfhIWkH&U8iQ89?I=EAyc4Nks#k%3T_(
zH?G*4&gcTSmDOii*2oH^o^w1m<8#YwMC!5%PfXCcb+7<ZDlv`YfOT6m@WZJC3`?Ig
z$;M1(Y0>qr%yc*BVzJGD79bBPjnBOMaxQNc-BIqsdaj{a{?fKnn{E9JXhs^YJ^;>K
z14KVrt80B$8duFa-g1^s($z0-Ge*uZkyu=_8ob{lTw)Jhdg?;ED<P{np1bvRFR$lZ
zwHcL|pDOUNtv|_^1*%sK2U^2tG(v;9XE8D%Yw9hN%k5?HYIjt<<e~kQ!G&H@sk=C1
z8uHcqpsTNs=Cw^Nx?i=om9Z(d2|1f<H{*wwknlyhG)HkW*H8CmsfX`UIHk@x_&>1+
zBxRsJu|{<4B-N#=$*aDl!w(7Bhfr5wfRXO~lcNSn_Abx~ngz-(<4ti*y586Rdz{Y0
z-p9CNyUq!?+%>EVUCbsNH)e@+FUKz(Z9J{2c@K(~9EzrZOUyhbOoo&F+wGIK84??j
zwCdqvdzzg2=|;UFKkgIf<C^987au9-o7}1Yxiq3?_)u>9XMjB<_<5Xa77sUn5`1nZ
zd?{*Gj1%u8GvJVs_Km{@5zJU!SRr45iCu&i=hbYfC)XiU<?6f--2u)8p19n~8dhOF
zH6@z1jaQ?(plIERZ(IOvfQa7y7k60Wb4=>??c1q*$7~0%^W;)RwU%mXZ)Lpd4tA%g
zDc;l;FBKrMm>m|@_T069r|~ZT>ENf6rK9sz`{Z-$|NP!zQpeV_6=H7$DJcqC$B-U^
z&Z$yKrI+vYzdRjaWnZJY^jUZxR8<k*vJwWHbv!mxzB#6Sto$*m<p_)sBF!Hi+TU}O
z?+#x_Q&ix8{ybGCl`%6qK^N>xzPg`X?8_vjSA&<SQ>A?y(9pw{fNgUc&Rv{9Y5zUq
zcu3W`Ec<y#y(rjJr1utIcsgx=*aNNYlO#mwdbh(2IO}o3Ywfrco+eNc0#qdnUYPtX
z`7g^Ry|l=qGi8wUsoPw#ihnNrLA#F3wn*j1h`&qGmkv%6i~RkK#_L1uaS6BC-Uv?s
z(I~BBdV=#Ia=dALCUf^^!*$<j2y=(IKy5ZP=YG<yXALfCGfa0ZuDtFTN-9PAV^_{1
z;~4{+zdo}fl;sNYp7QwBXA_Q#(5&TCTF6B`8%$L5jmO>5R5^A>^F7I;-ngIBQ-*)?
zy$Yo#DgK>Lq#g|cIg~KUyel&ZG*cZrpLpE2>^6aBY0&wGa9{UWp;pSoS<u!+LtUIw
z<3`60e4cUEtxMf9m$qZBhVTqb!W_FUJ_4Z=l1z<0rqCv)&NLW`CNIm0FDX3L`FTAQ
zA#@OKIqI54bPw6Don?66+yk9EA1lW=5iD?h!=}{o)PNBL>_i`byVfq5GP!roIs2vq
z?o}E2LFfpCRO|B%eq9ycY?Rd9*i%^sLa(_=W7?1fA<Pyk{wbGxM&)zgFFYNz5qL22
zXhL71yf?Nl1$dl69TzOz<AES$^%J=JvHvNMXr&u2&VtM#s=p?|F!DHsz7lN5@ZEM(
z^o;n-YYrdahq*Xz@ZIL8yg!S<08$A&#R({rM=NI0|FB=&X^14E6rB0xf}Y#X%s?yE
zD3FU}7BcGrqbJ8pCGF0hI5)S~s<-ZUy269AY?cf9`niQJsyptM)D};!hwilt%?9^?
zF=l{d?DNEZ4@#|&DS@J4+lBiP7j?7&)Xb}5x}v?mQCIGEF)uy6X3^{9BL|)7g;5@A
zz4U`Ank6<y1vwEDd9pOY>{7p{kr?4Ya*~Rt<*b{_^lQb2$(UEYiE)$HdY3k#MYj%g
z4bW$QZjl&!l^a^MjeCnqp<-n8lylD5ifDmb5Ny`ZGO6k&bhqd&i&yO}>NO|p?p+A%
z8d!#JLK2=@4|OnaoquCc_+Sh17W&cmbFdS5oj$Mrt8(MPwroIV(GWO?Xdgy@_pUSl
z0P0{={M<V|Wtz3L^Kf}CIw5|wOP&j7_Sew}!M`CaNX2XOF@I!|<6VIN1QFS;4^xoE
zm5q)Pp@P%j&N;sfsy@pQGiS^a<k^(3lXP;U;|3ca3xmGnbKRV?I07$qn5-k*?t&<I
z^8>s71N-VA&R%6<irpj*`3n1&Yd2&&W|;mtn}!55yLbalC@w5y&gjVUoi7yFPb(r@
zi@BdllRU^Johl}tH^Z<)8G%%eE8Sc&q&+dFk7KN@C{a3IrXLPzWc}rxxrCX%9-EuB
zYEEIxfo8STGwx8#H43(HAHJE`3k9Fg*zvs+%G=krIBRx}<K(@tDz)6m6n==S41_7g
zZpFw{+m8#but|x%lhjyqOQ8=oNBl1~h|9&x??$!SQ=XMtaeQ}@s#93A_WqBRfs{a5
z56Zca>573zO~F2x4hsf7Le=2{TP?XaY&F(&CLyG_gxmfzu6rjbdR1tDQ;s_Eevi*B
z_A=#fZM<8vx;430_-b*RQ?m*0b3MPS#k;RX85Z5WyVlOvAiA>VyZAbsgPF8j@|FF2
z;~^wGM!LJT+qX}kudPuecP2}c)}QM%?~!x7ETrflB*fiX=q}Z(Q6ZYwQ|Vq7wXHN`
z`Qqo=wlPs1mp#$%U=+ydoJ3UIvS~NuGQ%PC5@h5bb}CQ6DG)tdR?xrmvooGKw!JNG
zudzYsz7R?6jW1~eIAPAOYY_Ib$7h2*HopqYC0wBF(K>;I%1hnKvWJVyB22AknG&17
zhh)jUdV35%QU|rQ;s<2#-8G2?Do*9uyl&d%SO1>llG|%VcT}!g+E2B8!cK~7n~Zop
zkgM-O(i=O}nB1}@`dm~S6Gu&bNH_bX?%IPqEtP8yswbCile>fV+9jv1;&l#p<S|4y
zf_OqkOTd23y<npBit_J?@jth8>KyLxrvG8=&Kk^#91x?V)vkELvY98F^fI0~Z=QJ&
zS>&G;<KjeEAVVUhTsxYwz7RvApLw%28iQKF2&Uo`vsprB^O7tp!@cI+=v%KQ!}MKh
z_q5Wl-}OH6yHfjQC~8l}*j5(R@;`nTUw$ujj|VzJ1TuoxuKwDbQzY`oBu!|{E1pf>
zSn5+8GAO|RweI}S{Z`I2*oXLD<BJ|R$=5iy60ob!9j9@EVr4yVcT#fx>v}-ns(WrP
z_VU2#fcP9^I!`8)%xIp}zyT`v<{{H1R-lR_dwTHoD|igKfVk=6wWbfX>>ppjmH&L^
zPX_n@=k>`oZazmT<l&|u!LF7o3dOIo<kxIrKxH`+5_^?y@7!^LQYG;}@HH=PKntIA
zE%96i@mn$o$CjXIQ&z>*0omOtZODAT%UwP#3L+}_Dj0hBrX2hix@zIS%GT)W6Ogr8
z@<@e38(=P@b+?vdEm9D&ltR8!1nA9sZE6)R1xV2q(J?|tZ5&dZM9QL8=RmR;@#UPW
zGzk3#N&p5ST_ZN|k!aLB^wNew(pn4Hfdz2R(z&Ffs;e(RJQ5xL6~3;f7L@<Y(6hDh
zGs!Au$zwV_Z4^lUKoG~=hSEwy#<`*gKW<F{AEcC}u8br*P>BhKeg*mKwNPqF1{Pp^
z7vi4*NdCv&p;}`Ag{kq138AcvLIT%tXv_QxNrdQn7Swy)K<sD~qR1=oWf^JbHV~(k
zb2vHj4<xBD_5Ed`eR@zeBo?L37ymd(TD&_Uv>znJSt4rIoy`G}Gm%QjqaXVv@Q~pk
z%8W9b+6KXx1l0IshBuKWUI-s_NV<@mmR8SE)K7=V9ATg2K#oZdjeZe6$B60xOeDi~
zCX){Pi{%RsH33%V+Kk;Moy84VN0fRX8BFaggc7|1eDTjF<dxFZ>4sgyN;63{58?&g
zlOaR1sllQ6YEd-6bbY`s1PVq7EH#>bhd^^^#o&%*Jz)Tywo_VZ7`f*%yt7gjp(_nP
zHD(ao388ll#H%X}KOp)ahD<N&05z>ryTz^?04m`FOQzHUE*)i-FkYAW-dvx{c|YMx
zWUj=|(YBY#0Zq}sSN0;sQ44Y-j<SLdw>Zc@dn}94E=A_&9N1D<0f$7}*Nj>eX-C`5
z>UO_u>Myb7xPi`}9foH@!O*A*3UN{PNNZ92Ty;m89P}V&!dFc>3C)}>Q|>Z@Hfl#L
z5E8QAxLYV2FF2<~A<bKHV;ND$etiNp?jFBhUid|m?Ja0AOhJ1G+xl7KSVd<nn=0of
zAY~czRLd6o-jPW)Bt5K!1hC`voez1p5pl{)ch0&;*Y#f3DK$h`9?NHc%EGUM@mqO4
zkZ+dDu5DBOC3z({Iq^_pXF?>hlN_ELd4nEAjwK)rkWr7bd&rSy?5T~E8RdcPxOrON
z6AoOK7p^%Jdcu)*L(j_7ts3)jVm|c&QlQjsv@jd8rxcpmM!wTUT^ZCbq)tzQiw1I|
zFPzw2fF8oSZ}0RQ=N@oKfrdpoUa9PRjn$C6szJlWP@5;wMB7M#{v2>7FR-EWEpP!7
z%DiM2VdYQ+9g8xKTg=EQw00psv6P#)1ECf=S3KrlfdVE3lu;5q&T`O1vT~!-2AVc{
zkN?vav^x=F`y5}SA_~Vze<->+U#0%X*?ee>_te3870Rvn{#SpZ2(d1i8{XYnI2u1m
zThKCTdTnAuGk6Xb7zNvcF8fmnF=+a(JjFhQv38Bbb>X{Qeo}&f*^0voc6ST}N8COE
z`LleNZnqb0TSo6(ao6Iv?_38infwHgfWKvX-@38J`wJ62ZH8yT;U3ANXuo4$!E0}w
z0gWBGt9)RTjimTP$oA)!;M=GoHG0<)=$9)&a&fPFAL=tfAnuI}nnC)NphdhsfUEnF
zJwfv9sC$oLgrC6V*Ol!KS_SdWbzm1lJKK*bX<RNCQ%;NBf;_kJT2^zPRTT%)3ik~w
zQaV;Z@lF(V?#9Kku?9L`6=cCZzybc!Qh1A=njEUCnjNYdI?lfZJv)e?B$e^O(rB&G
zU_DT8(HcD<>b~R?@IGrB=@kZIAPf@6T97UwO^`+|^fK4WoOK|{GIB`++BRB6gE}->
zII+p_jr|h4i@)|vO8fPx#O2Uw%``so*G-Gg&_LLV>S<S>=51QPS~j*K^KUc4MXO+5
zSmTIUaD8hW#y5kQ9}jVhj_sL8W|zj*>;^z7lLBqNsh+9E?hZOt8zY3=BU&0f_M9^A
zxE1qgCb{08Mh%S{tKp^dXlA+6tv6Zay|>@&-V~^ZF)>`d+Swnxo=G?k?>o+9u_g?n
z6Y`C!xQW_1^c?6oym9hBUDRJlSMIc*h;y}6GRrD@pGP40p|-ExYW4o(E@9Ei`kuOx
zGON$ctyk(Ii4g6_CN}U4ZWA$QerdlDMxB?nnqz}%jM)~OSc1$Od|MR8ExtOzA9k{S
z3PRIJ_qC`3#HHZq<RiK5FiSOulQ^BjUvyzI3d4lqs^n3}Yc&Y#I)Z%vlK0XfT0u#f
zw!hd*&@$RH=p8%cA-yFK_EI4v{{SaJw+5VNYQJxPeLGVym_bkn1IOWA7~*ge|L-5p
b!aj;_jOpK>Z(Utc@JCWiRy12g_u>Bnvs1v0

literal 0
HcmV?d00001