parse multiple alleles into distinct Variant records

[iskandr]

Fixes #22.

Also fixes bug from the last PEP8 PR which removed reverse_complement from string_helpers.

[timodonnell]

I suppose this has always been the case for varcode, but this just made me think about it: we're throwing away the genotype information. Even for a regular non-multiallelic variant, I don't think I have a way to see if it was called a het or hom. Would it make sense to retain that in the variant? Or we could potentially address that in a later PR.

[iskandr]

Yep, currently we're only attaching the INFO dict to variants and not the sample info. What do you think is the right way to handle e.g. genotypes? Should we make a nested sample_name -> key -> value dictionary of dictionaries? That seems messy but I'm not sure what else to do. In any case, that should definitely be a separate PR.

[timodonnell]

LGTM