metastable defects

[alexsquires]

Currently, the code will not work (or at least not as intended) if a DefectSpecies has more than one DefectChargeState with the same charge. This doesn't allow us to consider metastable defects.

I think this issue is resolved as simply as moving away from having DefectChargeStates represented as a dictionary, and instead a list or array. Though there may be unforseen issues in the formalism.

[bjmorgan]

I can't think of a reason why the set of DefectChargeStates needs to be a dict and not some other collection type / or why it is necessary for the keys to be the charges (except that this is conceptually simple in cases where you do not have metastable defects).

[alexsquires]

I don't like the fact the keys are the charges anyway, it's redundant (I'm not exactly sure when that got worked in to the code). But fixing it would be a breaking change, this gives me a reason to do it.

[bjmorgan]

There are definitely places where that dictionary is iterated over to get keys or DefectChargeState objects, but these can hopefully be replaced with equivalent code in a fairly localised fashion.