feat: use tutor configs as django settings

[Henrrypg]

This pull request updates the way configuration settings for AI model integration are managed in the openedx_ai_extensions plugin. The main change is moving from hardcoded values in the backend to dynamic, configurable settings that can be managed through Tutor and environment variables. This makes the plugin more flexible and easier to configure for different environments.

Key changes include:

Configuration Management Improvements:

• Added new configuration variables (OPENEDX_AI_EXTENSIONS_VERSION, OPENEDX_AI_EXTENSIONS_API_KEY, OPENEDX_AI_EXTENSIONS_MODEL, OPENEDX_AI_EXTENSIONS_TEMPERATURE, OPENEDX_AI_EXTENSIONS_LLM_FUNCTION) to the Tutor patch file, enabling external configuration of AI model settings.
• Registered these new settings with default values in the plugin initialization (plugin.py), so they are available and can be overridden as needed.

Backend Refactoring:

• Updated the backend to use the new settings variables (OPENEDX_AI_EXTENSIONS_*) instead of hardcoded values for the AI model, API key, temperature, and LLM function in the workflow configuration.
• Removed the previous hardcoded assignment of AI model and API key from the plugin_settings function, making the backend rely solely on the externally provided settings.

[felipemontoya]

the 2 things I dont like (as per our call)

• too many variables, we can try something like what django does for DATABASES
• even though we are writing it tutor first, we should make it safe for someone to not use tutor, which it currently is not as the variables jump from tutor to backend code