Improve data augmentation functionality

[bdewilde]

Description

• Improved, extended, and added data augmentation transform functions
  • word-level transforms can now be limited to words of particular part(s) of speech
  • char-level transforms, broadly analogous to word-level, are now available for substituting, inserting, swapping, and deleting individual characters
• Added an Augmenter class to combine multiple transforms and randomly apply them to spaCy Docs in a variety of ways
• Refactored augmentation "utils" code, and changed a field on the AugTok named tuple
• Added tests for all of this

Motivation and Context

My first pass on data augmentation (PR #268) was okay, but woefully insufficient. This brings the sub-package up to a passable level.

How Has This Been Tested?

Wrote a bunch of tests and ran the code locally, and everything checks out.

Types of changes

• [x] Bug fix (non-breaking change which fixes an issue)
• [x] New feature (non-breaking change which adds functionality)
• [ ] Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

• [x] My code follows the code style of this project.
• [TODO] My change requires a change to the documentation, and I have updated it accordingly.