Add ambiguous genotypes in VCF output

[johnlees]

Currently any non-ACGT IUPAC codes are written as an 'N'. It would be possible to keep the information by writing out multiple output genotypes for a sample, with equal probabilities.

Not yet sure whether this will be a much used feature, and would take a little effort to implement.

[johnlees]

Not going to do this unless a few people ask for it