Parse patches on a per-language basis

[damevski]

Short term goals:

• tokenize changesets (for each lang.)
• remove stop words (for each lang.)

and insert into Solr

[damevski]

We could add a few regular expressions specific to each language here. For instance, removing variable names if they are on the left side of an '=' sign in most imperative languages. It would help control the noise.

[rohvani]

Updated issue title to be more accurate, #15 ties into this.

[damevski]

Use SrcML for now