feat(providers): add Rapid-MLX provider

[raullenchai]

Summary

Adds Rapid-MLX as a new local
provider. Rapid-MLX is an OpenAI-compatible inference server for Apple
Silicon, built on mlx-lm.

• Install: pip install rapid-mlx
• Run: rapid-mlx serve <model>
• Endpoint: http://127.0.0.1:8000/v1
• API: OpenAI chat completions, streaming, tool calling
• Repo: https://github.com/raullenchai/Rapid-MLX

Files

• providers/rapid-mlx/provider.toml — provider entry, mirrors the
  existing lmstudio provider (uses @ai-sdk/openai-compatible).
• providers/rapid-mlx/models/... — a small set of representative
  MLX-quantized models. More can be added incrementally.

Notes

• Pattern follows the existing lmstudio provider. No code changes.
• Model IDs match the exact strings users pass to rapid-mlx serve.
• Tool calling is supported via Rapid-MLX's parser layer.