adjust gather to save to disk after every iteration

[ctb]

it's really annoying to have no idea where the job is when watching a big gather run on a remote node via snakemake 😆 . if we flush to disk, we could probably see things more quickly and it would make me happier.

we could also do something like iteratively output the matching signatures, which would let us restart the job more quickly.

Maybe add a flag --batch or --long-running or something.

[ctb]

(or just --save-frequently)

[ctb]

this is something that prefetch in #1370 could do quite easily.

[ctb]

#1493 actually supports progressive output of signatures during gather if you use a directory or a .zip file!! So, I think I'm going to close this. but tagging in #1350 just in case.