other neighbor list options

[rkurchin]

not high-priority, but I am curious how sensitive accuracy is to these since I did find that there was a noticeable different between the Ulissi group's Voronoi approach vs. the more "naive" cutoff distance (surprisingly, the latter seemed to work better, at least in the cases I tested)

see ideas here: https://pymatgen.org/pymatgen.analysis.local_env.html (source here)