potential next steps for the sourmash sketch command

[ctb]

#1159 adds sourmash sketch dna|protein|translate. This issue tracks some potential additions to that --

• we could provide additional subcommands, like sourmash sketch reads, sourmash sketch genome, and sourmash sketch ncbi, that choose good defaults for those kinds of inputs
  • e.g. for reads, scaled=1000,abund
• we are removing 10x support so no need to add anything there.
• similarly, we could alias sourmash sketch dayhoff and sourmash sketch hp.
• could add sourmash sketch 16s (see e.g. classification with full length 16S gene #1000)

Additional thoughts:

• @luizirber proposed a passthrough option, where you could chain the inputs together:
  • sourmash sketch dna --passthrough <input> | sourmash sketch protein <input> - this would yield two signatures. Not sure how to do the output tho - where does the signature file go? do we need to specify multiple different -o options?!
• translate should take options to specify top-strand/rc only, for RNA.
• can we make the sketch module suitable for direct use by Python, per Idea: imports match CLI? #1112?
  • e.g. sourmash.sketch.dna(filenames=...)
• the sketch module should auto-recognize signature files and not try to load them as FASTA... sourmash compute doesn't recognize signature files #814
• might be good to add automated traverse functionality to discover all FASTA files, if possible, at least for specific commands...

[ctb]

I think maybe the stuff in my comment here #999 (comment) around detecting/flagging bad inputs should be examined for sourmash sketch to see if we've resolved that UX disaster...

[ctb]

maybe add name to param strings per #1315