feat: enhance reasoning tooling UI

[qnixsynapse]

Problem Statement

In the current implementation, the tool calling and reasoning UI displays each reasoning and tool invocation step as separate entries in a stacked list.
For models that perform extensive reasoning or make frequent tool calls, this results in an overly long and cluttered thread, making it difficult to navigate or review.

Feature Idea

Introduce a collapsible “reasoning & tooling” thread to streamline the interface.
Each reasoning or tool call sequence can be collapsed or expanded, allowing users to focus on key outcomes while still having the option to inspect detailed reasoning when needed.

Expected Benefits

• Cleaner, more navigable UI for complex reasoning chains

• Improved user focus on key reasoning results

• Scalable design for models with multi step reasoning activity

[github-actions]

---

Found 2 possible duplicate issues:

1. #4496
2. #4620

---

[qnixsynapse]

Initial attempt:

Problem: After tool calling, the response received is a different response
Possible consideration: attempt using responses like API.