fixed bug in which F1R2 and F2R1 annotation kept discarded alleles

[davidbenjamin]

@LeeTL1220 you were right. Fortunately the fix is very simple.

[codecov-io]

Codecov Report

Merging #3636 into master will decrease coverage by 0.259%.
The diff coverage is 100%.

@@              Coverage Diff               @@
##             master     #3636       +/-   ##
==============================================
- Coverage     79.78%   79.521%   -0.259%     
- Complexity    18237     18606      +369     
==============================================
  Files          1225      1232        +7     
  Lines         66983     68558     +1575     
  Branches      10449     10884      +435     
==============================================
+ Hits          53439     54518     +1079     
- Misses         9320      9732      +412     
- Partials       4224      4308       +84

Impacted Files	Coverage Δ	Complexity Δ
...bender/tools/walkers/annotator/OxoGReadCounts.java	86.139% <100%> (-9.514%)	58 <2> (+39)
...institute/hellbender/utils/read/SamComparison.java	57.403% <0%> (-22.961%)	86% <0%> (ø)
.../walkers/contamination/CalculateContamination.java	89.552% <0%> (-1.827%)	37% <0%> (+17%)
...der/tools/walkers/contamination/PileupSummary.java	90.196% <0%> (-0.98%)	24% <0%> (+11%)
...trics/analysis/BaseDistributionByCycleMetrics.java	100% <0%> (ø)	1% <0%> (?)
...oadinstitute/hellbender/tools/GatherVcfsCloud.java	70.811% <0%> (ø)	40% <0%> (?)
...tute/hellbender/tools/AnnotatePairOrientation.java	96.429% <0%> (ø)	8% <0%> (?)
...nder/cmdline/PicardCommandLineProgramExecutor.java	57.143% <0%> (ø)	2% <0%> (?)
...ols/walkers/markduplicates/MarkDuplicatesGATK.java	89.837% <0%> (ø)	76% <0%> (?)
... and 15 more

[LeeTL1220]

Do you think it would be prudent to add/enhance a test that just checks that the number of values is correct? I

[LeeTL1220]

@davidbenjamin Can you make a test for this?

[davidbenjamin]

@LeeTL1220 done. I put it in the M2 integration test because it is part of a bigger potential issue, that of phantom alleles from the likelihoods that don't make it into the variant call. Back to you.

[LeeTL1220]

@davidbenjamin Looks good... merge away