Dialect prototyping

[elijah-potter]

Issues

#345

Description

This is a large patch, so there's a lot to cover:

• Created a Dialect type to represent the 4 major English dialects
• Refactored WordMetadata, Token, and TokenKind to no longer be Copy.
  This affected performance by about -8%.
• Reworked the MutableDictionary (and by extension the FstDictionary) to use word hashes as keys in a single map.
• Added a derived_from element to the WordMetadata. This allows linters to determine the base word (the base word for bananas is banana) for an affixed word.
• Renamed the hunspell module to rune. This module will slowly evolve to container a more bespoke dictionary file format. It has already evolved from the hunspell format sufficiently for a name change.
• Assertions error messages have been improved.
• Made it possible to construct a MutableDictionary at runtime using Rune files.
• Imported words from MIT-licensed dictionaries for en-GB, en-CA and en-AU.
• Added dialect selectors to the Obsidian and WordPress plugins
• Added a search bar for Harper rules in the Obsidian plugin.

Next Steps Before Merge

Before I can feel comfortable merging this PR, there are a couple things that have to happen:

• More non-American versions of words must be added to the dictionary and tagged as such. See the existing diff for the dictionary for an idea of how this works.
• Rune needs to be extended to be able to describe the relationships between words of various dialects (color -> colour).
• Additional unit and integration tests must be added for the newly-supported dialects.

Edit: the built-in spell check seems to convert between dialects quite well on its own. There isn't any immediate need for manual tagging.