Add support for MonteCarloMembraneBarostat in Sire-to-OpenMM

[lohedges]

We have ported the Cresset MonteCarloMembraneBarostat implementation to SOMD1, but there is now the need to test this with SOMD2. As such, we'll need to add support for the MonteCarloMembraneBarostat to the Sire-to-OpenMM conversion code, along with an option via Sire dynamics.

The current SOMD1 implementation is fairly limited and makes some strict assumptions about the box setup. As such, as a first-pass implementation it might make sense to provide a hidden option to enable the MonteCarloMembraneBarostat via the map, rather than a direct kwarg. That way we can test the implementation then later choose the best way to expose it via the API, i.e. what kwargs would be needed. (The SOMD1 implementation assume zero surface tension, XY isotropic, and Z free, which isn't what we'd want for a general implementation.

[mark-mackey-cresset]

XY isotropic and Z free is what nearly everyone is going to be using, as that's the format chosen for the OPM database which is what everyone uses to align membrane proteins. The surface tension is something that people might want to adjust, though.

[lohedges]

Thanks for the clarification. I plan to work on this as soon as I get some spare time.

[mark-mackey-cresset]

Thanks Lester!

[lohedges]

No problem. Apologies for not doing this sooner. The surface_tension dynamics option has been exposed in somd2 on my feature_gcmc branch. I plan to merge this into main soon.