OpenBabel-produced PDB files lack TER records

[avirshup]

It appears OpenBabel doesn't write TER records in PDB files.

This can become a problem with tleap specifically, which detects terminal residues via
a) TER records or
b) the end of a chain.

So, if a chain includes a bunch of residues AND some waters or a ligand, we can't parameterize it.

Options:

1. alleviate (but not eliminate) this by reading PDBs from mmCIFs, which split chains more aggressively.
2. manually write out TER records
3. switch to another PDB writer
4. Split structures into chains within MDT, at least before sending them to something that needs TER records

[avirshup]

Option 1 didn't even work - the mmCIFs provided by PDB have problems and biopython parses them strangely.

Option 2 - seems unnecessarily hacky, but in practice should be very straightforward - we just need to insert TERs at the ends of lines.

For option 3 - what PDB writer? Biopython doesn't do bonds, meaning we'd have to do CONECT records manually. RDKit doesn't accept tetravalent nitrogens, suggesting that it's not flexible enough for our use cases.

Option 4 is appealing, but probably introduces too much unnecessary confusion - users will see chains appear out of nowhere.