Implement chat templates for LlamaLanguageModel

[mattt]

No description provided.

[Copilot]

Pull request overview

This PR implements chat template support for LlamaLanguageModel, enabling proper multi-turn conversation formatting using the model's embedded chat templates (e.g., Llama 3, Mistral, ChatML). Instead of using raw prompt text, the model now applies conversation-aware formatting by extracting the full transcript history and applying the GGUF-embedded chat template via llama.cpp's llama_chat_apply_template API.

Key Changes:

• Adds formatPrompt(for:) method to convert transcript entries into formatted prompts using model-embedded chat templates
• Adds extractText(from:) helper to extract text content from transcript segments
• Updates both respond and streamResponse methods to use formatted prompts instead of raw prompt descriptions

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.