Supporting multi-language conversations

[gunthercox]

Right now, the format of the data organizes the dialog by language. It would be useful if the format of the corpus supported the ability to annotate individual statements to specify a language.

---

Add a default_locale attribute to corpus files: "Primary language for a conversation".
Individual statements should be able to support languages of different parts. (Including start and end index). A validator will be needed for testing.

[gunthercox]

Possible option to explore:

Add an optional languages sub-attribute to text to permit language-specific annotations.

Rough example:

- "Language A, Lagrange b."
  - "Language B response"
  # define languages with sub-string indexes
  languaes [A: [0:12], B:13:24]