Add Custom Model Support and Configurable API URL

[DDnim]

Why

• Enable users to use newer or custom OpenAI or compatibility models that are not pre-defined in the plugin
• Allow users to configure custom API endpoints for compatibility with OpenAI-compatible APIs
• Skip token counting for custom models since their context limits may vary

What

Added Features

1. New "customize" option in model selection dropdown

   • When selected, shows an additional input field for custom model name
   • Users can input any valid model identifier (e.g., gpt-4-1106-vision-preview)

2. Made OpenAI API URL configurable

   • Moved hardcoded URL to a configurable setting
   • Maintains the default OpenAI endpoint as fallback
   • Allows users to use alternative API endpoints

Technical Changes

1. Added CUSTOMIZE option to CHAT_MODELS
2. Added customModelName field to settings
3. Modified token counting logic to skip for custom models
4. Updated settings UI to dynamically show/hide custom model input
5. Enhanced type safety and error handling
6. Maintained backward compatibility with existing configurations

Testing

• Custom model name input appears when "customize" is selected
• Token counting is skipped for custom models
• Default OpenAI URL is preserved when no custom URL is set
• All TypeScript type checks pass
• Worked for next models / endpoints
  • grok-beta(https://api.x.ai/v1/chat/completions)
  • gemini-flash-1.5(https://generativelanguage.googleapis.com/v1beta/openai/chat/completions)

[DDnim]

@rpggio
Hi friend.
I really like this plugin and hope to use more models available.
Please review it when you have time, THX.