[DNM] Prototype using pablo to load pdb files

[jthorton]

This is a quick investigation into using pablo to load pdb files for the protein components.

Notes

• due to openff topology having a nice interface to openmm topology the changes needed to load pdb files are minimal
• comparing the topology created by loading from openmm vs pablo are different (this just seems to be extra information like bond orders etc are added by Pablo)
• pdbx and mmcif are not supported yet
• Some tests fail, we should look into if this is a pablo issue or an issue with the files
• we should make sure we can load all of the industry benchmark systems with pablo to have parity with the current loader.

#Questions

Checklist

• Added a news entry

Developers certificate of origin

• I certify that this contribution is covered by the MIT License here and the Developer Certificate of Origin at https://developercertificate.org/.

[IAlibay]

Some initial questions:

• How is the performance? If you load in a large protein, do you get any noticeable differences in performance?
• What dependencies does Pablo require? Do we need to pull the entire OpenFF toolkit? (including interchange?)
• Is atom and residue order preserved?
• Does it roundtrip well?
• How easy is it to pass through new definitions for things that aren't in the inside CCD dictionary?
• Does it behave well when you have a specific protonation state? For example protonated GLU.
• How does it recognise CYS bonds? Will it attempt to find them automatically (like. PDBFile) or does it require them to be explicitly defined?