Unpractical to pass a dictionary to CrystalNNFingerprint when aiming at augmenting with chemical information

[nisse3000]

Referring here to a comment by @computron in PR #250:

"In practice, I think people are going to have a hard time passing in the chem_info dict and find this to be a pain to use. While flexible, normally you have a dataset of 1000 compounds and don't want to manually construct a dictionary of all possible species to some value like electronegativity.

It might be better if this was able to easily work with an "AbstractData" class instead, since those already have a bunch of built-in dictionaries for various properties. Or, some easy way for the user to leverage those classes rather than construct the dictionary themselves.

Or, maybe instead AbstractData itself could implement a method that turns its properties into a dictionary. That way you could do PymatgenData.dict_output("X") that will turn all the PymatgenData electronegativities to a dictionary format."