Improvement: Add a Local Model Manager to application

As a user, I'd like to be able to use the application, and have it load a model I have already downloaded previously.

The issue describes adding a local model manager to a browser-based LLM chatbot. Start by examining the application's model loading and storage logic, likely in the main TypeScript source files. Look for existing model handling code and browser storage APIs. Determine how to detect previously downloaded models and integrate a management UI. Testing will involve verifying models can be loaded locally without a server.

Written by the indexing model from the issue text.