use picklists for ignoring identifiers instead of custom coding it in the Snakefile

[ctb]

#255 and #259 added skip_genomes in the config file, and added code in the Snakefile to handle it.

I think this was short-sighted - where genomes need to be ignored, they should be ignored from the very beginning, in the prefetch stage. This will allow sourmash to pick the best available genome.

I'm not 100% sure where to do this - we already support inclusive picklists in genome-grist, and FBFW we don't support multiple picklists in sourmash (either via the CLI or internally for many database types) - but will figure it out!