added new prompt approach, audio input (in different languages - whisper), colab

[ejri]

WebLM_interactive is a different approach to the webLM approach crawler approach. It's still in the works, with a lot of room for improvement and optimizations.

It takes user input and breaks the request into a list of tasks, which would potentially make it easier/reduces the number of steps to complete the request.

Additional features: It allows users to:

• use audio input in addition to text input.
• allows users to have their audio input in any language

WebLM_interactive and weblem would eventually merge into one weblm, combining all features from both. I've added a colab notebook to follow some of the steps. (the notebooks needs a bit of cleaning up too)