Add tags attribute to models

[gunthercox]

This change is going to be a prerequisite to making ChatterBot's response selection process more efficient. Using categorized training data from the chatterbot corpus, it will be possible to do text classification on input statements to place them in a group based on the existing training data. Once tagged, a strong assumption can be made by the chat bot that the appropriate response to the input statement exists within the set of statements that have the same tag, or category as the input statement.

Because filtering the results in the database based on a named attribute can be made to be fast and efficient, this will help solve a large portion of the performance problems that the current version of ChatterBot faces (~0.7).

Note: this pull request will only add the tag attributes to statements. Future pull requests will be required to:

1. Allow storage adapters to filter results based on tags
2. Modify the the training processes to add tags based on the:
   a. Categories for data in the ChatterBot corpus
   b. File names in the Ubuntu Dialog corpus