Improve the Interface

[phuang00]

• update dependency_parsing.py and dunning.py to reflect the class based structure that has been implemented in the other analysis modules
  - Should we convert dependency_parsing.py to work for a corpus instead of just individual documents?
  - Currently, creating dependency tree takes a very long time. Would using Standford's updated stanza package improve the efficiency?
  - For dunning.py, should we have users input a corpus and a filter instead of having users create two corpuses?
• Add doctests to improve coverage
• Erica mentioned how the toolkit was originally used on a binary structure (male or female). Check to ensure that everything is updated to work with new structure of more than 2 genders.