ENH: Handle other periodic boundary conditions

[HaoZeke]

eOn depends on the assumption that all the external potentials (and some of the output writers) use the same definition of PBCs for wrapping. However, this is not always true.

After discussions with Hannes, the cleanest way for now is to just maintain all the varying implementations (there aren't that many [1,2,3]), and allow the Potentials to pass the type as a parameter.

Refs:
[1] Deiters, U. K. Minimum Image Convention Coding of Microcomputers. Molecular Simulation 1989, 3 (5–6), 343–344. https://doi.org/10.1080/08927028908031386.
[2] Hloucha, M.; Deiters, U. K. Fast Coding of the Minimum Image Convention. Molecular Simulation 1998, 20 (4), 239–244. https://doi.org/10.1080/08927029808024180.
[3] Deiters, U. K. Efficient Coding of the Minimum Image Convention. Zeitschrift für Physikalische Chemie 2013, 227 (2–3), 345–352. https://doi.org/10.1524/zpch.2013.0311.

[HaoZeke]

Worth keeping MDAnalysis/mdanalysis#1475 in mind as well.

In general the current approach in eOn is to use fractional coordinates, rescale and then back-transform, which is very solid for orthogonal boxes (if a little expensive), not so sure about triclinic.