Use augur subsample

[victorlin]

This workflow's subsampling is implemented by custom rules that call augur filter one or two times and combines the results. It works fine, but complicates the Snakemake workflow DAG, is ad-hoc, and is not portable across pathogen repos.

Description of work

1. Update the configuration from:

   rsv/config/configfile.yaml

   Lines 18 to 42 in 170f99c

   	filter:
   	group_by: "year country"
   	min_coverage:
   	genome: 0.3
   	G: 0.3
   	F: 0.3
   	min_length:
   	genome: 10000
   	G: 600
   	F: 1200
   	resolutions:
   	all-time:
   	min_date: 1975-01-01
   	6y:
   	min_date: 6Y
   	background_min_date: 1975-01-01
   	3y:
   	min_date: 3Y
   	background_min_date: 1975-01-01
   	subsample_max_sequences:
   	genome: 3000
   	G: 3000
   	F: 3000

   to something more generic and customizable.

2. Replace the filter_recent, filter_background, and combine_samples rules with a rule that calls augur subsample.

   This rule should:

   1. Allow concurrent sample runs with threads
   2. Pass a dump of Snakemake's config as a YAML file to --config
   3. Extract the relevant configuration using --config-section
   4. Allow Snakemake to intelligently handle conditional runs of the subsample rule in the case of config changes.