Allow choice of sample when running multi-tumor patients in paired mode with a single normal

[yashpatel6]

          > The deletion doesn't actually change any behavior at all in the metapipeline; the metapipeline already greedily takes the first normal BAM and uses it while ignoring the others when running multi-tumor patients in paired mode. Also, the deletion is only happening under /scratch, the final output directory still contains all of the normal BAMs generated in their respective output directories.

The change here is purely to clean up /scratch by removing files that are already not being used further in the metapipeline

Got it. End users might want to rerun somatic pipelines retrospectively if the specific normal sample has potential issues, especially when the normal processed with mets samples was used while primary one is available. It would be useful to eventually have an option to choose which normal sample to use.

Originally posted by @tyamaguchi-ucla in #139 (comment)

[yashpatel6]

          Agreed, I figure it hasn't been a huge issue so far since this only applies when run with a specific combination of samples and run mode; if multi-tumor samples are run in multi mode or if the normal-tumor samples are explicitly paired and then run in paired mode, this behavior should have no effect. I'll add an issue for allowing users to choose for that specific case though!

Originally posted by @yashpatel6 in #139 (comment)