Subsampler and augur workflow

[dpark01]

This adds a WDL task for the subsampler package, which subsamples genomic data proportional to externally provided epidemiological case counts. This exposes it in a new standalone WDL workflow called subsample_by_casecounts and adds a new workflow augur_from_msa_with_subsampler which glues the case-count-based subsampling step to the standard augur / nextstrain pipeline.

The new task deconstructs subsampler's provided Snakemake pipeline and calls the four scripts manually, in part because the Snakefile doesn't have a good way of passing input parameters at runtime, doesn't seem to honor all of those input parameters within its own rules, forced mandatory parameters that were actually optional, and had a few other bugs to sort out.

This has been tested internally in a few runs on Terra and succeeds at reducing the entirety of the ncov/open dataset to a manageable size for downstream augur steps.