Remove contamination

[leoisl]

As we are trying to keep a low memory footprint and want to be able to distribute as many resources as possible with tbpore, we only really need to use the H37Rv reference and maybe a few other NTMs as well. Then we keep anything that maps to H37Rv.

Some NTM references can be found in https://www.nature.com/articles/srep45258#Sec25 and https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0214088

The code I used to do this in the head to head was https://github.com/mbhall88/head_to_head_pipeline/blob/3ed818c9f3754612a33060428ee651cd0f91234b/data/QC/Snakefile#L272-L383 (ignore the rule for extracting illumina reads obviously)

Original issue: #1

[leoisl]

To be implemented in v0.2