Improve a bunch of things

Full Changelog: v1.0.0...v1.3.0

Fine I'll make a changelog

• Add message sanitation
• Store conversation history
• attempt to fix the typing notification
• add CPU inference requirements
• make gaslighting actually work
• try and fail at zeroshot response characteristics(stuff like [no-response] and [reply to @ whatever]
• comment out debug-only code(on multiple occasions)(yes I debug with print statements)
• add this cool thing where the bot randomly replies to you 5% of the time. Also add a bunch of other randomized response characteristics, such as multi-line responses by sampling the number of responses from some weird random distribution, more natural responses by needing to only respond with "you" rather than the bot's full name, etc
• try to log stuff to text files(there's nothing to log)
• some random refactoring sprinkled here and there
• fix a few bugs caused by my own incompetence
• attempt to run a larger model using parallelformers and fail
• parse mentions in messages so that the bot doesn't get confused

That's basically about it

AMOGUS HAS ARRIVED

This works so insanely well that I decided to make a release. I'm honestly surprised I got it working this well in just 6 hours

Full Changelog: https://github.com/sagarreddypatil/nlp-chatbot/commits/v1.0.0