feat: implement dynamic model fetching from GitHub Models API with pr…

[LoackyBit]

This pull request adds support for dynamically fetching and caching available GitHub-hosted models, and updates the default model selection to use the latest GitHub model. The main changes introduce a new utility for fetching model metadata from GitHub, integrate this into the application's startup flow, and update model selection logic to prefer dynamically fetched models when available.

GitHub Model Fetching and Integration:

• Added a new utility module githubModels.ts to fetch available models from the GitHub Models API, cache results, and provide a prefetch mechanism. This includes logic to read authentication tokens and handle timeouts and errors gracefully.
• Updated the startup logic in src/main.tsx to prefetch GitHub models alongside Ollama models, ensuring model metadata is available early in the app lifecycle. [1] [2]
• Modified the model options logic in modelOptions.ts to prefer dynamically fetched GitHub models over the static list, allowing users to select from up-to-date model offerings.

Default Model Selection:

• Changed the default GitHub provider model from 'github:copilot' to 'gpt-4o' to reflect the latest recommended model.