SelectVariants use a GATKPathSpecifier for output

[jean-philippe-martin]

Instead of a String previously.

Tested like this:

$ ./gatk SelectVariants \
--variant src/test/resources/large/VQSR/dbsnp_132_b37.leftAligned.20.1M-10M.vcf \
--select-random-fraction 0.01 \
--output gs://$BUCKET/variants.vcf

and

$ ./gatk SelectVariants \
--variant src/test/resources/large/VQSR/dbsnp_132_b37.leftAligned.20.1M-10M.vcf \
--select-random-fraction 0.01 \
--output /tmp/variants.vcf

If this works, the plan is to change all tools to use this format, so they can all write VCF output to Google Cloud Storage.

[codecov-io]

Codecov Report

Merging #5870 into master will increase coverage by 0.002%.
The diff coverage is 100%.

@@              Coverage Diff               @@
##             master     #5870       +/-   ##
==============================================
+ Coverage     86.84%   86.842%   +0.002%     
  Complexity    32274     32274               
==============================================
  Files          1987      1987               
  Lines        149001    149001               
  Branches      16469     16469               
==============================================
+ Hits         129392    129395        +3     
+ Misses        13596     13594        -2     
+ Partials       6013      6012        -1

Impacted Files	Coverage Δ	Complexity Δ
...org/broadinstitute/hellbender/engine/GATKTool.java	91.204% <ø> (ø)	102 <0> (ø)	⬇️
...der/tools/walkers/variantutils/SelectVariants.java	80.914% <100%> (ø)	131 <0> (ø)	⬇️
...e/hellbender/engine/spark/SparkContextFactory.java	71.233% <0%> (-2.74%)	11% <0%> (ø)
...nder/utils/runtime/StreamingProcessController.java	67.773% <0%> (+0.474%)	33% <0%> (ø)	⬇️
...utils/smithwaterman/SmithWatermanIntelAligner.java	90% <0%> (+10%)	3% <0%> (ø)	⬇️
...ithwaterman/SmithWatermanIntelAlignerUnitTest.java	90% <0%> (+30%)	2% <0%> (ø)	⬇️

[cmnbroad]

Since we already have the bucket test for this code path from the last PR, looks good.