Solvate with non-cubic boxes

[IAlibay]

trafficstars

OpenMM now supports non-cubic boxes, we should see if we could pass this as a user input (through SolventComponent?).

[IAlibay]

pinging @hannahbaumann

[richardjgowers]

As a sanity check, could we first quickly check for a sample set of protein structures (PLB?) how much solvent we can avoid adding and as a post-step quickly check that the benchmarking figures are agreeing with this being quicker.

I'm also curious if aligning the protein's gyration tensor along box primary vectors would help here

[jameseastwood]

Can we get a review on this (really the associated PR) sometime soon? @IAlibay @hannahbaumann Low priority though. [] Work out if it makes sense [] Try it out on a few systems