Packaging force field files along with Protocols

[IAlibay]

Following up from today's alchemiscale call - it would be good to consider if / how we can package in force field files from host machines to compute ones.

This will particularly be important when using bespoke force fields.

[dotsdl]

From discussion:

• having openmmforcefields process a URL given as the FF; would then download and use
• the FF itself as a long string given as the FF

From OpenFF: probably want functionality in about 3 months From ASAP: probably want within ~1 month

[jthorton]

Hi ! Just pinning this PR here as a possible solution to the problem, I have found that with this change the entire bespoke force field string can be set as the small molecule force field at gufe.settings.models.OpenMMSystemGeneratorSettings.small_molecule_forcefield Onece merged we could sanity test this by running the TYK2 benchmark set again with openff-2.0.0 but set as a string as follows:

from openff.toolkit import ForceField
from gufe.settings import OpenMMSystemGeneratorFFSettings

sage = ForceField("openff-2.0.0.offxml")
ff_settings = OpenMMSystemGeneratorFFSettings()
ff_settings.small_molecule_forcefield = sage.to_string()

[ijpulidos]

This is a cool quick approach to the problem, thanks for working on this @jthorton

[dotsdl]

Is there anything remaining to close this now that openmmforcefields 0.12.0 is out?

[IAlibay]

I believe this still needs discussion - we will need this when we go towards gmx, files will be serialised then and we'll need to work out how to ship itps.