Move subsampling by proximity into augur

[sidneymbell]

Context
The ability to subsample using a focal set and then background context selected by genetic relatedness is SUPER useful. We'd like to be able to do this for other pathogens.

Description
Move the subsampling logic that assigns proximity scores by genetic relatedness out of ncov and into augur proper.

Examples

Possible solution

1. From James on Slack: Add a command like augur utils genetic-proximity {focal_seqs} {contextual_seqs} {n-per-focal-seq} {output.strains.txt} and then if you need a sequences/metadata output then you can pass the strains.txt to augur filter.

[huddlej]

@jameshadfield has implemented our subsampling logic as a script in ncov, but we haven't revisited this with more testing. We originally planned to implement this as augur subsample (and @jameshadfield created a corresponding Augur PR), but it seemed likely at the time that we'd substantially change this subsampling logic and we didn't want to codify the current implementation in augur only to change it again.

Would you all be up for trying out James's script version of subsampling for SARS-CoV-2 or other pathogens? That could help us figure out if it's worth the effort to move into Augur...

[sidneymbell]

Oh awesome, thanks @huddlej and @jameshadfield! We'd be up for trying out the script and potentially helping to move things over to augur if helpful, although we probably won't be working on this until a bit later / early spring.