Add a tracing server configuration e.g. Phoenix Tracing

[lasendo]

I would like to have the following feature:

Within the configuration settings, let it be the YAML file or other place, I would like to be able to specify the URL of a Phoenix tracer server (or other LLM observability tool, e.g. MLFlow), and instrument it so that all LLM interactions are saved.

This will serve as a means to persist the LLM call data, and use other functionality from the tracing tool, such as labelling answers, etc.

I see this specially useful if the project name can be passed.

This way, if the tracing server is set, then all the teamates can tap into what is being accomplished by LLMs.

Thank you for the consideration!