mqtt: Improve handling of multiple PDU parsing

[jlucovsky]

Continuation of #10236

Issue: 6592

Link to redmine ticket: 6592

Describe changes:

• Parse PDU instead of entire stream

Updates:

• Reflect remaining byte count (per review comment)
• s-v test pr update

Provide values to any of the below to override the defaults.

To use a pull request use a branch name like pr/N where N is the
pull request number.

Alternatively, SV_BRANCH may also be a link to an
OISF/suricata-verify pull-request.

SV_REPO=
SV_BRANCH=pr/1609
SU_REPO=
SU_BRANCH=
LIBHTP_REPO=
LIBHTP_BRANCH=

[suricata-qa]

Information: QA ran without warnings.

Pipeline 17783

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (c3b3c11) 82.28% compared to head (a3612eb) 82.32%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #10262      +/-   ##
==========================================
+ Coverage   82.28%   82.32%   +0.03%     
==========================================
  Files         977      977              
  Lines      271950   271950              
==========================================
+ Hits       223784   223885     +101     
+ Misses      48166    48065     -101     

Flag	Coverage Δ
fuzzcorpus	63.52% <100.00%> (+0.12%)	⬆️
suricata-verify	61.50% <100.00%> (-0.03%)	⬇️
unittests	62.82% <0.00%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

[glongo]

According to Victor's suggestion (#10058 (comment)):

the frame should be created earlier, and not depend on the parser returning ok. It should be created as soon as we start parsing the data, and then be updated with the length when it's complete

For reference, you can look here for reference:

7366506#diff-d988869639f4f6ed35c7047322b592a33a7421d22c38aa6137d456967b3656d3R162-R200

[catenacyber]

Should it not be current.len()-rem.len() ? (and same below)

[jlucovsky]

Will adjust in new pr

[jlucovsky]

Continued in #10279