Support for using 'langchain serve' endpoints

[GustavFredrikson]

• New LangChainServeLLMGenerator Class: Implements HTTP POST request handling to interact with LangChain Serve LLM APIs, allowing integration with Garak probes.

[github-actions]

DCO Assistant Lite bot All contributors have signed the DCO ✍️ ✅

[GustavFredrikson]

I have read the DCO Document and I hereby sign the DCO

[GustavFredrikson]

recheck

[leondz]

Thanks for this! We'll get it reviewed

[leondz]

Thanks for this, looking pretty good

• We need to fix how model_name is used
• Config variables need to be handled cleaning
• Exception handling can get clarified a bit - interested in your feedback re: intended behavior with this generator & when it might make sense to backoff, skip, or abort
• What do you think we can do in regard to tests?

[GustavFredrikson]

• Implemented URI validation to ensure the API endpoint is correctly formatted before attempting to connect.
• Replaced model_name usage with an environment variable (LANGCHAIN_SERVE_URI) for setting the API endpoint, let me know if other usage is preferred!
• Error handling:
  • Logging for different types of HTTP errors (4xx and 5xx).
  • Changed return values from empty strings to None.
• Removed unused config params i.e. temperature and penalties that are managed server-side.
• Added tests covering URI validation, initialization, mocked generation, and error handling scenarios using requests_mock.

[jmartin-tech]

Some minor thought and I think this will be ready to go!