Create sphinx documentation for simplify, join, insert, and powerset

[hmhummel]

Also, added equivalency of each parameter to Tikka's causaleffect functions. Added detail to parameter documentation.

[djinnome]

Hey @hmhummel Sphinx will display the mathematical formulas such as $P(J|D)$ and $P(V|C) := P(V_k |C_k)$ in latex if you put a :math: before the backquotes, for example:

:math:`P(V|C) := P(V_k | C_k)`

Note that I included underscores in Vk and Ck so that they would be subscripted.

[hmhummel]

Oh, awesome!! Love this & will implement in the next commit — thanks

[hmhummel]

Documentation for causaleffectpy

I began to document causaleffectpy functions in a similar fashion to causaleffect so that this can eventually be integrated into y0 and used as a reference to write the probabilistic simplification algorithms.

I used sphinx to document 4 functions in .rst files, and used this sphinx tutorial with the make html command to generate html files.

Unit testing for causaleffectpy

Unit tests for all functions still need to be created with pytest using the 3 vignettes/test cases from causal.effect on CRAN: 1st with unobserved confounders, 2nd with pruning & fractions, and 3rd with simplest graph possible (no unobserved confounders). For reference, here are the causal.effect unit test files using the testthat R package: test_case_1, test_case_2, and test_case_3. @djinnome recommended using causaleffectpy to help create the unit tests rather than y0.

Needs work:

• simplify - created sphinx documentation and html file. Changed names of functions to y0 but not sure if they're right. Needs an example/unit test
• join - created sphinx documentation and html file. Changed names of functions to y0 but not sure if they're right. Needs an example/unit test
• insert - created sphinx documentation and html file. Changed names of functions to y0 but not sure if they're right. Needs an example/unit test
• powerset - created sphinx documentation and html file. No related y0 functions.
• deconstruct
• causal.effect — might not be needed, as there is already an identification algorithm written for y0
• dSep
• wrap.dSep
• parse.expression
• factorize
• probability
• get.expression
• etc