Protein RFE: Define a suitable prootein-protein atom mapping object

[IAlibay]

One of the current "hurdles" to doing protein RFEs is that we haven't decided on the right AtomMapping object for protein-protein transformations.

Could we just use LigandAtomMapping for now and do the giant int:int dictionary optimization later?

TODO:

• Rename LigandAtomMapping?
• Add necessary convenience / docs to get protein-protein atom mappings out of Kartograf?

[mikemhenry]

I vote for that, until it falls over that will be the most powerful way (since we can specify the mapping entirely) and that means a non-protein away atom mapper can be coerced to work as well