Aiosqlite to mongo

[zavocc]

Using MongoDB means faster feature updates, ease of maintenance, DOCUMENT/JSON based so its easier to work with it and use native types like dict instead of separate SQL queries and scalability

TODO:

• [x] Test async (NEEDS IMPROVEMENT)
• [x] Use of index? and various optimizations (STILL LEARNING THIS)
• [x] Some code organization and refactoring

NEEDS TO ADDRESS:

• [x] Fix initial response times as it is slow esp on async calls of ask command, due to database connection? Or HistoryManagement instantiation in Init? In this case, learn more about motor (WTF?)

[zavocc]

Refactored tools, cogs and updated some docs TODO: testing and review

[zavocc]

New features in this version so far:

• Move from SQLite to MongoDB using Motor
  • CHAT_HISTORY_DB is now MONGO_DB_URL (connection string)
• Add show_stats parameter in /ask command to show the chat conversation turn counts and the models used (off by default)
• Deprecate JSON mode, users must instruct the model to format its responses to JSON
• Refactoring some code
  • Function calling code is reduced / Tool definitions and functions are refactored to simply import them as modules and parse Tool class and its properties instead of using separate tool/function registration and use _callable syntax and mandatory convention to call functions.
    • When calling tools, the interstitial now displays a human readable name instead of the function name
  • Inherit / merge chat_mgmt.py and genai.py to simply chat.py (controls) which inherits from generative.py (LLM generation) so the generative module is now responsible for database connection and LLM lifecycle. It is still a cog/discord command where it contains /ask
  • Update docs (WIP)
  • prompt_history is now a counter for how many chat interactions was made instead of counting arrays arrays of prompt which is a privacy implication if one views the database to check chat length based on turns