improve validation for membrane systems

The only thing I just realized is that if the user provides a fully solvated protein-membrane system using the --protein flag, it will create the transformations without an error. This is also not handled in the API, but I wonder if it's even more dangerous on the CLI level. One could e.g. also add a density check on the ProteinComponent and if it's above a threshold, but not a SolvatePDBComponent, at least add a warning? Though people probably don't look at warnings since there are so many. This is probably something that should be handled elsewhere but still wanted to raise it here. Do you have any thoughts @IAlibay ?

Couple of takes here (already communicated in earlier meeting, just writing it down here):

• It's probably ok to go with this release without that check.
• Adding some check for the component that sees if you have both box vectors and high density would be good. My initial take is that a warning would be better, but I do agree that we just have too many warnings. We could either: 1) give validate an optional "error on warning" that is used by more sensitive tooling like the CLI, or 2) just go with the error and see how many folks it affects.

Originally posted by @IAlibay in https://github.com/OpenFreeEnergy/openfe/issues/1896#issuecomment-4325857292

No files or tests are named. Start by tracing validation for the --protein CLI path and the API, then inspect how ProteinComponent inputs with box vectors and high density are handled. First resolve whether this should be an error or warning; done means the chosen behavior consistently catches fully solvated protein-membrane systems in both entry points.