Handle NA variants

[CNuge]

Currently, ./. are given a status of unknown, but there are also instances where NA values may make their way into the variant files due to other issues with the genotypes.

We need to decide if we want the burden of cleaning up the genotypes to be on the user (i.e. impute REF, go back to the read counts and hard code a gt, other user-side solution) or if we want to have silent means of handling missing data (either skipping these individuals or imputing REF).

I'm inclined to throw a verbose error and tell the user to fill in the data; but thinking forward, this will not play nicely if we ever wanted to apply the method to a whole vcf (with messy info).

[CNuge]

Now addressed with #14