Contextualspellcheck

This release fixes bug(30). Extra changes:
- setup staleBot
- converted TODO task to issues
- setup discussions

Thanks nicno90 for raising the bug!

This release
- fixes 26.
- improved the code quality by enforcing style consistency (23).

Thanks dc-aichara for the PR on flake8 and thanks nicno90 for pointing out a bug in the package!

Note: Had to yank release v0.3.1 because of debug print statement in production package.

This release adds following important changes:
1. Feature addition: Specifying the transformer model
This will enable the package to be **used on other languages** as well for which `transformers` has a model
2. Feature addition: specify the maximum edit distance
3. Update: Vocab loading now happens from the transformers tokenizer. Previously it used a static file (vocab.txt)

A big thanks to dc-aichara for **contributing Feature addition 1 and 2**.

This release specify the core dependency of the project in the setup. This was suggested (17) and added by AlvaroCavalcante in 18.

Following changes in this release:
1. Allow user specified vocab to identify misspell (12)
2. Fix a bug where file read presented issue. Thanks to gaurav0804 for 11

Thanks mark-fangzhou-xie for pointing out an issue where the token text and position is modified an a downstream process. This affected our custom extension.

As a fix
- use setter for corrected sentence and not getter (to solve issue with downstream process)
- raise warning when extension is accessed and the token position/text is modified