Refactor MT wdl to make validations easier

[meganshand]

This pulls the bulk of the pipeline into a separate subworkflow so that the validations (with the mixture samples) can be run. The mixtures have already been subset and tagged, which is why the rest of the pipeline needed to be extracted.

[codecov-io]

Codecov Report

Merging #5708 into master will increase coverage by 0.018%.
The diff coverage is n/a.

@@               Coverage Diff               @@
##              master     #5708       +/-   ##
===============================================
+ Coverage     87.069%   87.086%   +0.018%     
- Complexity     31874     32034      +160     
===============================================
  Files           1940      1940               
  Lines         146739    147271      +532     
  Branches       16226     16380      +154     
===============================================
+ Hits          127764    128253      +489     
- Misses         13061     13077       +16     
- Partials        5914      5941       +27

Impacted Files	Coverage Δ	Complexity Δ
...utectReadThreadingAssemblerArgumentCollection.java	66.667% <0%> (-3.333%)	2% <0%> (ø)
...ller/ReadThreadingAssemblerArgumentCollection.java	92.593% <0%> (-1.852%)	1% <0%> (ø)
...allerReadThreadingAssemblerArgumentCollection.java	70.588% <0%> (-0.84%)	4% <0%> (+1%)
...der/tools/walkers/mutect/M2ArgumentCollection.java	89.231% <0%> (-0.243%)	23% <0%> (+9%)
...r/tools/walkers/mutect/Mutect2IntegrationTest.java	92.188% <0%> (+0.194%)	108% <0%> (+16%)	⬆️
...lotypecaller/readthreading/ReadThreadingGraph.java	89.753% <0%> (+0.782%)	226% <0%> (+67%)	⬆️
...ypecaller/AssemblyBasedCallerGenotypingEngine.java	91.228% <0%> (+1.316%)	91% <0%> (+1%)	⬆️
...ller/readthreading/ReadThreadingGraphUnitTest.java	97.136% <0%> (+1.652%)	90% <0%> (+31%)	⬆️
...pecaller/readthreading/ReadThreadingAssembler.java	69.755% <0%> (+1.678%)	85% <0%> (+33%)	⬆️
...ithwaterman/SmithWatermanIntelAlignerUnitTest.java	90% <0%> (+30%)	2% <0%> (ø)	⬇️
... and 1 more

[ldgauthier]

Looks reasonable. Since you're still running samples I trust that any wiring mishaps will pop out.