Integrate AG2's Custom Logging for Runtime Data Collection

[dwhan89]

Description:
To collect more runtime data and to help us debug, I suggest that we use AG2's custom logging framework to collect runtime data. This implementation will allow us to monitor the CMBAgent's behavior and provide a full runtime record of where the agents got it right and wrong. Eventually, we may even be able to use the logs to fine-tune the agents.

---

Proposal

1. Integrate AG2's custom logging system into the project.
2. Configure the logging framework to capture runtime data, such as:
   • Full decision tree stack.
   • Whether the run was successful or not.
3. Make sure that logs are structured, consistent, and saved to a designated folder.
4. Provide configuration options for log verbosity and output destinations (e.g., console, files, or external logging systems).

---

Goals

• Improves visibility into the model's runtime behavior.
• Simplifies the debugging process by capturing detailed logs of errors and events.

---

Misc Notes

1. We could implement this ourselves using the Python standard library, but it seems like AG2 team has already been implemented it (https://docs.ag2.ai/docs/reference/runtime_logging). So, it’s probably just a matter of using this feature throughout and also making it configurable at runtime.

[dwhan89]

Note: Another non-critical feature. I will go through the codebase and see how this can fit in.