feat(opencode): add auto loading models for litellm providers

[farukcankaya]

What does this PR do?

Fixes #13891

This PR adds auto loading models feature for LiteLLM proxy providers in OpenCode. When a provider is configured with both litellmProxy: true and autoload: true options, OpenCode will automatically fetch available models from the LiteLLM proxy's /model/info endpoint (with fallback to /models endpoint). This eliminates the need to manually define every model in opencode.json when using LiteLLM proxies that may have many models.

How did you verify your code works?

• Unit tests
• Manual testing e.g. for opencode.json

  "$schema": "https://opencode.ai/config.json",
  "model": "InternalLiteLLMGateway/anthropic/claude-opus-4-5",
  "provider": {
    "InternalLiteLLMGateway": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "InternalLiteLLMGateway",
      "options": {
        "baseURL": "https://litellm.internal",
        "litellmProxy": true,
        "autoload": true,
        "apiKey": "sk-xxx",
        "headers": {
          "Authorization": "Bearer sk-xxx"
        }
      }
    }
  }
}

litellm-auto-model-loader.mp4