Add support for writing energy components to file and preserving the LJ sigma parameter for ghost atoms

[lohedges]

This PR adds support for writing the energy components (from each force) to a log file for debugging purposes with the --save-energy-components command-line option. This also builds on OpenBioSim/sire#198 and OpenBioSim/biosimspace#295 to prevent perturbation of the LJ sigma parameter for ghost atoms, which causes instability during perturbation.

[mb2055]

Looks good, problematic systems seem to run fine now 👍

[lohedges]

I've updated this to match the logic from Sire, i.e. all zero sigmas are set to the opposite end state value, not just those for ghost atoms. I've also included a fix for @jmichel80's issue, where it isn't possible to create a system from a perturbation template when the free system only included the ligand and water. This makes sure that the search returns a SelectorMol iterable, so the existing loop logic still works.