Implement PathSeq taxon hit scoring in Spark

[mwalker174]

Upgrades PathSeqScoreSpark to perform abundance score calculations on the executors rather than the driver. This was crashing on inputs with a lot of pathogen reads.

This also required some minor changes to the PSPathogenTaxonScore class to be able to keep track of abundance score contributions that come directly from hits to that taxon and those that are from the taxon's descendents.

As a result, some of the test output changed when using bitwise, exact checks on the output. So the tests now check for output equivalence, meaning parsing the scores table, checking that all the taxa are the same, and that the scores are equal to within some defined epsilon.

[codecov-io]

Codecov Report

Merging #3406 into master will increase coverage by 0.093%.
The diff coverage is 89.474%.

@@               Coverage Diff               @@
##              master     #3406       +/-   ##
===============================================
+ Coverage     80.497%   80.589%   +0.093%     
- Complexity     17553     17668      +115     
===============================================
  Files           1175      1175               
  Lines          63487     63836      +349     
  Branches        9895      9963       +68     
===============================================
+ Hits           51105     51445      +340     
- Misses          8433      8434        +1     
- Partials        3949      3957        +8

Impacted Files	Coverage Δ	Complexity Δ
...nder/tools/spark/pathseq/PSPathogenTaxonScore.java	88.235% <87.5%> (-11.765%)	3 <2> (+1)
...itute/hellbender/tools/spark/pathseq/PSScorer.java	88.77% <90.244%> (-0.825%)	60 <20> (+6)
...tionbiasvariantfilter/OrientationBiasFilterer.java	95.492% <0%> (-0.064%)	80% <0%> (+24%)
...ools/spark/pathseq/PSFilterArgumentCollection.java	80% <0%> (ø)	3% <0%> (+1%)	⬆️
...bender/tools/walkers/annotator/FragmentLength.java	100% <0%> (ø)	8% <0%> (+2%)	⬆️
...e/hellbender/utils/variant/GATKVCFHeaderLines.java	99.315% <0%> (+0.029%)	7% <0%> (ø)	⬇️
...itute/hellbender/tools/spark/pathseq/PSFilter.java	93.567% <0%> (+0.95%)	33% <0%> (ø)	⬇️
...e/hellbender/engine/spark/SparkContextFactory.java	71.296% <0%> (+1.296%)	18% <0%> (+8%)	⬆️
...bender/tools/walkers/annotator/OxoGReadCounts.java	94.828% <0%> (+1.494%)	28% <0%> (+11%)	⬆️
...tools/walkers/genotyper/AlleleSubsettingUtils.java	90.062% <0%> (+1.88%)	55% <0%> (+13%)	⬆️
... and 7 more

[tedsharpe]

On vacation--won't have a chance to look at this until next week. If Chris or others approve just go for it.

[cwhelan]

This looks fine to me.

[cwhelan]

Git -> Get?

[mwalker174]

Done