Added maxRetries to runtime in M2 WDLs

[davidbenjamin]

@takutosato This is really important for overcoming transient errors in Firecloud, especially as long as we don't have call-caching for NIO.

[LeeTL1220]

@davidbenjamin @takutosato Can you default to zero? Unless you can think of another use case, do we need this once we leverage the working NIO call caching?

[takutosato]

Looks good. I don't have a preference for default being 0 or 3. Up to you two @davidbenjamin @LeeTL1220

[LeeTL1220]

If no other use case than the NIO workaround can be identified, I worry that you will do a bunch of retries on a terminal condition. Setting the default to zero would protect against that.

[codecov-io]

Codecov Report

Merging #5049 into master will increase coverage by 0.019%.
The diff coverage is n/a.

@@               Coverage Diff               @@
##              master     #5049       +/-   ##
===============================================
+ Coverage     86.359%   86.377%   +0.019%     
- Complexity     28809     28853       +44     
===============================================
  Files           1789      1791        +2     
  Lines         133512    133725      +213     
  Branches       14914     14936       +22     
===============================================
+ Hits          115299    115508      +209     
- Misses         12811     12815        +4     
  Partials        5402      5402

Impacted Files	Coverage Δ	Complexity Δ
...ools/copynumber/formats/records/LegacySegment.java	46.875% <0%> (ø)	7% <0%> (?)
...r/formats/collections/LegacySegmentCollection.java	72.222% <0%> (ø)	4% <0%> (?)
...ute/hellbender/tools/copynumber/ModelSegments.java	99.654% <0%> (+1.25%)	65% <0%> (+21%)	⬆️
...tools/copynumber/ModelSegmentsIntegrationTest.java	98.5% <0%> (+9.397%)	30% <0%> (+10%)	⬆️
...ithwaterman/SmithWatermanIntelAlignerUnitTest.java	90% <0%> (+30%)	2% <0%> (ø)	⬇️
...utils/smithwaterman/SmithWatermanIntelAligner.java	90% <0%> (+40%)	3% <0%> (+2%)	⬆️