Prune assembly graph before checking for cycles

[davidbenjamin]

@ldgauthier This doesn't hurt M2 on our evaluations but should I do anything to check that it's okay for HaplotypeCaller?

[ldgauthier]

M2 is definitely more sensitive, so if you're happy I'm happy.

I'm also happy (and surprised) that none of the integration test outputs change, but I hate changing code without changing tests. Do you think the user would let us use a hundred or so bp of his/her data as a regression test?

[ldgauthier]

Actually, what about Sarah's MT data instead? That should be fine to use.

[codecov-io]

Codecov Report

Merging #5562 into master will decrease coverage by 1.404%.
The diff coverage is 9.091%.

@@               Coverage Diff               @@
##              master     #5562       +/-   ##
===============================================
- Coverage     87.062%   85.658%   -1.404%     
+ Complexity     31879     31243      -636     
===============================================
  Files           1940      1940               
  Lines         146799    146748       -51     
  Branches       16233     16227        -6     
===============================================
- Hits          127806    125701     -2105     
- Misses         13073     15050     +1977     
- Partials        5920      5997       +77

Impacted Files	Coverage Δ	Complexity Δ
...aplotypecaller/HaplotypeCallerIntegrationTest.java	5.388% <0%> (-82.718%)	9 <0> (-76)
...pecaller/readthreading/ReadThreadingAssembler.java	63.077% <100%> (-5%)	47 <0> (-5)
...ct/CreateSomaticPanelOfNormalsIntegrationTest.java	3.448% <0%> (-94.253%)	2% <0%> (-8%)
...ls/walkers/mutect/CreateSomaticPanelOfNormals.java	0% <0%> (-93.846%)	0% <0%> (-21%)
...r/tools/walkers/mutect/Mutect2IntegrationTest.java	5.111% <0%> (-86.882%)	2% <0%> (-90%)
...alkers/mutect/SomaticReferenceConfidenceModel.java	12.5% <0%> (-84.375%)	1% <0%> (-7%)
...ls/genomicsdb/GenomicsDBImportIntegrationTest.java	9.313% <0%> (-79.157%)	2% <0%> (-78%)
...kers/haplotypecaller/ReferenceConfidenceUtils.java	0% <0%> (-66.667%)	0% <0%> (-1%)
...der/tools/HaplotypeCallerSparkIntegrationTest.java	2.439% <0%> (-65.854%)	2% <0%> (-14%)
... and 75 more

[davidbenjamin]

@ldgauthier I put in a regression test from the user's data.

[ldgauthier]

Beautiful, thank you!