Fix issue #312

[lohedges]

This PR fixes #312 by skipping RMSD alignment when merging monatomic ions with other molecules, e.g. water when creating an alchemical ion. When a single item mapping is detected and one of the molecules has a single atom, then we skip the alignment and simply replace the coordinates of the mapped atom with those of the reference. A similar approach could be applied in the mapping/merging code with Sire.

(Note that there is some code duplication below since we still have two scoring functions, one for RDKit mappings, and one for Sire.)

• I confirm that I have merged the latest version of devel into this branch before issuing this pull request (e.g. by running git pull origin devel): [y]
• I confirm that I have permission to release this code under the GPL3 license: [y]

Suggested reviewers:

@chryswoods

[lohedges]

Bah, it looks like all of the AMBER related tests on Linux have inexplicably started failing. It appears that the sander.LES executable is being used, so perhaps this isn't working as expected, or behaves differently for the same input. Will try to debug locally. Amazing that this happens two days after release.

FAILED tests/Process/test_amber.py::test_minimise[backbone] - AssertionError: assert not True
 +  where True = <bound method Process.isError of BioSimSpace.Process.Amber(<BioSimSpace.System: nMolecules=631>, BioSimSpace.Protocol.Minimisation(steps=100, restraint='backbone', force_constant=10.0 kcal_per_mol/angstrom**2), exe='$PREFIX/bin/sander.LES', name='test', work_dir='/tmp/tmp5cpac605', seed=None)>()

[lohedges]

Nope, it's the exact same version of ambertools, which matches the version that I have locally:

    ambertools:                       23.6-cuda_None_nompi_py310ha112c55_105 conda-forge

Everything is fine locally, and for the builds the other day. If a single Python variant failed then I'd think that it was a freak occurrence, but all of them did 🤷‍♂️

[lohedges]

Looks like the issue is that sander.LES is the first executable with sander that it's detecting during the CI. I've excluded LES from that match. Fingers crossed it works.

[lohedges]

Yes, that did the trick. Really confused why this just started happening. Would have been nice if it happened before the release, since it's possible that people creating a fresh environment will experience the problem.