CBSS: normalize encoding of language field

[wlpotter]

Required to implement srophe/Gaddel#154.

First step is to pull this data out into a CSV for profiling and cleanup.

Possible solutions (TBD):

• Include only ISO codes in the language field (two-letter preferred, comma-separated, ordered by main lang than other langs)
• Allow a "LanguageNote" or "LanguageComment" field in the extra field that contains a prose description as needed