feat: implement GET /v1/models endpoint

[kianwoon]

Problem

Claude Code (and other clients) call GET /v1/models on startup to discover available models. Currently this returns 404 Not Found.

$ curl http://localhost:3456/v1/models
404 Not Found

Expected Behavior

Return a list of available models from the configured modelRouting and provider definitions, e.g.:

{
  "object": "list",
  "data": [
    { "id": "MiniMax-M2.7", "object": "model", "created": 0, "owned_by": "minimax" },
    { "id": "glm-5.1", "object": "model", "created": 0, "owned_by": "glm" },
    { "id": "qwen/qwen3.6-plus-preview:free", "object": "model", "created": 0, "owned_by": "openrouter" }
  ]
}

Notes

• Low priority — core proxy functionality works fine without it
• Not blocking any current workflows
• Will revisit later (label: enhancement, low-priority)