Fix overlapping fragments

[dmx2]

There are various protein tree entries that having fragments overlapping in residues. This is because UniProt can include multiple segments of a protein that overlap and currently we extract everything in the API call. Leidos doesn't have a way to parse this, I don't think, since if an epitope is contained in the overlapping range, it will show both.

First, we should see how frequently the fragments we provide are overlapping (in counts and %), then come up with some rules on how to avoid this during the Arborist build.

[dmx2]

Implemented:

Rules

• Only applicable to multiple, overlapping fragments
• If two chains differ by certain number of residues:
  • If only by one methionine, keep longest
  • If two chains or more, one full-length vs "cleaved" (or other names) indicate in the name
• If a propeptide, signal peptide, etc... overlaps a chain, keep them both
• If chains have a name in UniProt API, use it