RNASum glue should fail if multiple WTS samples for the same subject

[alexiswl]

We shouldn't assume pairings of DNA and RNA when there exists multiple.

Same goes for multiple DNA tumor normal analysis with different libraries, given an incoming WTS sample, we don't know which umccrise/sash DNA analysis to use as inputs

[victorskl]

Portal used to do the latest picking strategy. ditto

... use latest-greatest strategy

i.e. It will use most recent umccrise and transcriptome outputs of recently sequenced library, respectively.

Having said. IMO, it is perfectly ok to make a pause / graceful exit - assume we gonna re-introduce LaunchPad. Plus, user need define TCGA dataset anyway.