feat: add support for observability dashboard like LangFuse or LangSmith

Would be nice to have out-of-the-box support for LLM-specific observability platforms.

One that I particularly like is LangFuse, which has built-in functionalities to score and evaluate reasoning trajectories and agent performance, e.g., using LLM-as-a-judge.

Jaeger is basically for tracing only, which is fine but is only one part of the story.

What do the folks think?