Bug: LLM switches back to default in chatbotAgent after model change

## Bug Description
After switching the model in chatbotAgent, the LLM switches back to the default LLM when a new message is typed by the user.

## Suspected Cause
This could be due to the set_initial_model node being recalled whenever the user types a new message, which is not required.

## Suggested Fix
- Investigate the logic in chatbotAgent where set_initial_model is triggered.
- Consider adding a condition to prevent set_initial_model from being called after the initial selection.
- Look for alternative workarounds if simply updating the logic is not feasible.

---
Please update the logic or implement a workaround so that the selected model persists across user messages unless explicitly changed by the user.

Start in chatbotAgent and trace where the set_initial_model node is triggered when a user sends a new message. Reproduce the model switch followed by another message, then verify that the selected model persists across messages and changes only when explicitly requested. No specific file or test is named, so locate the relevant entry point and existing coverage first.

Written by the indexing model from the issue text.