Make VoiceStar pip installable, add Python API, update CLI + Gradio demo

[fakerybakery]

Hi,

Thanks so much for releasing VoiceStar - the audio samples sound incredible!

This PR contains a few changes:

1. Makes it installable w/ pip (so you can install it and run the CLI w/o having to clone the repo)
2. Adds a Python API for ease of use (so other libraries can more easily integrate it)
3. Automatically downloads models from Hugging Face (removes the need to run wget manually, caches models to prevent multiple downloads)
4. Add mps support for Apple Silicon
5. Updated CLI + Gradio demo
   a. Updated CLI to use click for better argument parsing
   b. Updated Gradio demo to download weights, use new API

Thanks!

(Sorry for the poor code quality in some parts, this PR is still very much a WIP)