Elmd

Added vector_to_formula for inverse featurizing

Added the fast EMD implementation, accessible through metric="fast". This is based on the method described here: https://arxiv.org/pdf/1804.01947.pdf

Merges in the optimimizations made by dwiddo

Adds a `full_feature_vector()` method to the `ElMD()` class which returns a complete featurization vector (n=8076) for each composition. Returns the mean, weighted mean, min, max, range, and std. deviation of all available elemental feature lookup tables (excluding permutations of one hot encoded atomic scales).

Fixes pip bugs that occurred in development for versions 0.4.16-18 due to mismatching version numbers in setup.py and __init__.py

Added full_featurize() to give a complete desciptor utilizing all featurizing dictionaries. Returned features take the weighted mean, mean, min, max, range, std deviation of the featurized elements in each composition