feat(providers): add Groq provider (`internal/providers/groq`)

[Nithwin]

Goal

Enable ultra-fast inference using Groq API endpoints (https://api.groq.com/openai/v1) for open-source models (e.g., llama-3.3-70b-versatile).

Requirements

1. Create internal/providers/groq/provider.go implementing ai.Provider.
2. Adapt OpenAI-compatible endpoint payloads specifically for Groq's high-speed API.
3. Register via ai.Register("groq", New) inside init().