Rigid Body Only Respects X-Ray Term

[kmdalton]

This is more of a design choice than a bug, but I wanted to document it here. The RigidBodyRefinementStep ignores non X-ray terms in the refinement target. This makes some sense in that it only refines the xyz coordinates and in a way which shouldn't impact the stereochemistry. Even if the ADP and/or stereochemical restraint terms entered into the objective their gradients would be zero. However, one can imagine scenarios where rigid body should respect custom coordinate losses. For instance, if you had harmonic restraints on the location of a chain in a large complex, you might want this restraint to be observed by rigid body. I can see this being useful at low resolution and especially if you have a partial model. That being said, I don't imagine this a huge priority. More something to keep in mind if there is a future rewrite.

[HatPdotS]

So on dev, it creates a completely new loss state now. This should be pretty easy to interact with if you wanted to add other loss terms. This is, I think, the design pattern I want: just create a new loss state rather than edit and roll back. It seems more straightforward to me.