Change to use low-level api where possible and reduce fetch calls

[keiranmraine]

Used 5000 WXS calls for benchmark, results are identical.

Final speed up 324s -> 150s (-53.7%).

Main items and how they were modified:

• CavemanPostProcessor::PostProcessor::_evaluatePentamerCheck:
  • Switch to low-level fetch
  • Restructure to only call fetch once per position
  • Removed some unused params and loops around commented print statements
• CavemanPostProcessor::_callbackTumFetch and CavemanPostProcessor::_callbackMatchedNormFetch
  • Removed redundant flag tests
  • Switch to low level API, only converting to AlignWrapper if useful read
  • only decode the individual base qual that is needed instead of the whole string
  • changed string split and access by index to be simple substr.

Original (following feature/strandCountFix):

Revised code result:

Main items and how they were modified:

• CavemanPostProcessor::PostProcessor::_evaluatePentamerCheck:
  • Switch to low-level fetch
  • Restructure to only call fetch once per position
  • Removed some unused params and loops around commented print statements
• CavemanPostProcessor::_callbackTumFetch and CavemanPostProcessor::_callbackMatchedNormFetch
  • Removed redundant flag tests
  • Switch to low level API, only converting to AlignWrapper if useful read
  • only decode the individual base qual that is needed instead of the whole string
  • changed string split and access by index to be simple substr.

[keiranmraine]

Internal CI completed, no difference to result from the feature/strandCountBug branch