Refactor sort_star_input.py

[adthrasher]

          I guess it's fine for now, but can you open an issue for refactoring this? I think the best set up would probably be rename and move `sort_star_input.py` into the `util` image, and then do the STAR input checking as part of `parse_input` instead of as part of the STAR task? I think that would also remove the need to host our own STAR image, wouldn't it?

Originally posted by @a-frantz in #139 (comment)

• Rename the script and move it to a different container (util?).
• Refactor the STAR task to remove the use of the script.
• Add the script call to parse-input at the RNA-Seq workflow-level.
• Remove our star Docker image in favor of clean image from biocontainers

[a-frantz]

Complication: currently, sort_star_input.py is fine with the RGs and FASTQs being out of order (e.g. R1s: [rg1.fq, rg2.fq, rg3.fq], R2s: [rg2.fq, rg3.fq, rg1.fq] RGs: [rg3, rg1, rg2]) because it will sort them in the output files. That would be messy to parse back into WDL, but we could instead have the script fail on bad orderings.

Point being, a bad order to rnaseq-standard is acceptable and recoverable, but will go undetected (and maybe lead to strange behavior?) in the Hi-C workflow