add phonology module

[thatbudakguy]

we need to be able to represent things like phonemes as data and transform them to/from vectors. this should help #21 and underlies most other work.

#16 is a prereq for assigning this data to Docs (and retrieving it at the span/token level)

• store features of phonemes as data — prereq for support tagging phonological features #7
• translate individual phonemes to/from numpy ndarray
• compare phonemes via vector distance (probably cosine) — prereq for create phonological scorer #2
• group phonemes into logical units (cluster, syllable)

[thatbudakguy]

see https://github.com/dmort27/panphon