Downsample mito bams to prevent eklipse crashes

[alkc]

Mito genomes w/ coverages reaching 40-60k crash eklipse, in contrast our QC guidelines mandate a minimum mean coverage of 500x, which is well under 40000-60000x

We should probably downsample to maybe 5000x, or even lower?

This update would most likely require a rerun of the mito validation samples.