Add SO type to HGNC genes

[kevinschaper]

Recently I wanted to update one of our slides that depends on counting the number of protein coding genes, and was surprised to see that we don't know have the type field populated for human genes.

One approach would be to use an explicit SO term from a different hgnc source (like alliance), I don't think we have a reasonable way to capture provenance on a field level, but we could definitely do a koza map to make the alliance hgnc gene record available alongside the one that comes directly from hgnc, and look up the SO term within the alliance doc.

The other approach would be to map from hgnc_complete_set.txt. That's pyobo's approach:

https://github.com/biopragmatics/pyobo/blob/dc7b4736f2bbf943084e8f8a95e1293c2717c566/src/pyobo/sources/hgnc.py#L110-L145

We could even consider importing this, though, it might depend on how heavy the dependency ends up being.