Optimized parseGenotypes routine

[cmdcolin]

At PAG a user gave me an idea for plotting many genotypes at once on the VCF track.

This is something that was done in jbrowse 1 with multivariantviewer, but it requires iterating through the genotypes of many features quickly. Ideally, we would want to see e.g. 1000 genomes data from a zoom level of e.g. 500kb

We could optimize the parseGenotypes routine e.g. the "get SAMPLES" routine

here is an example of how it looks parsing

2,800 variants, taking 8 seconds from a 100kb zoom level at 1000 genome data