Use instructor for structured output and make llama.cpp compatible.

[medihack]

No description provided.

[Copilot]

Pull Request Overview

Updates the chat client to use the instructor library for structured LLM outputs and ensures compatibility with llama.cpp by enabling Jinja templating and updating Docker image tags.

• Replaced OpenAI's beta structured output API with instructor library for more flexible LLM compatibility
• Added llama.cpp Jinja templating support and updated Docker images to latest versions
• Simplified the data extraction flow by removing intermediate parsing steps

Reviewed Changes

Copilot reviewed 3 out of 4 changed files in this pull request and generated 1 comment.

File	Description
radis/chats/utils/chat_client.py	Refactored to use instructor library instead of OpenAI's beta API for structured outputs
pyproject.toml	Added instructor library dependency
docker-compose.dev.yml	Enabled Jinja templating and updated llama.cpp Docker image versions

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

The removal of the assertion check for the parsed event could lead to returning None or invalid results. The original code had assert event to ensure a valid response was received before returning it.

Suggested change

	logger.debug("Received from LLM: %s", result)
	logger.debug("Received from LLM: %s", result)
	assert result is not None, "LLM returned None for extract_data"