[Question]: Agent Loop mode configuration not persisted across task runs in v2.1.0 console

# QwenPaw Bug Report

## QwenPaw Version

v2.1.0

## Description

In the console (frontend web UI), after modifying the Loop mode of an agent and running a conversation/task, the Loop mode automatically reverts to the default "Default" mode upon completion, rather than preserving the user-selected mode (e.g., "Goal" or "Task").

Expected behavior: The user-selected Loop mode should persist throughout the session or task execution and remain unchanged until manually modified again.

**Related PR(s):** #(optional)

**Security considerations:** None

## Component(s) Affected

- [ ] Core / Backend (app, agents, config, providers, utils, local_models)
- [x] Console (frontend web UI)
- [ ] Channels (DingTalk, Feishu, QQ, Discord, iMessage, etc.)
- [ ] Skills
- [ ] CLI
- [ ] Documentation (website)
- [ ] Tests
- [ ] CI/CD
- [ ] Scripts / Deploy

## Environment

- **QwenPaw version:** v2.1.0
- **OS:** [Please specify: macOS / Ubuntu / Windows version]
- **Install method:** Docker Compose
- **Python version (if applicable):** 3.12

## Steps to Reproduce

1. In the console frontend UI, navigate to the agent configuration page.
2. Change the Loop mode from "Default" to "Goal" or "Task".
3. Start the agent to execute a conversation or task.
4. After the task finishes running, check the Loop mode setting in the UI.

## Actual vs Expected

- **Actual:** After the task completes, the Loop mode automatically reverts to "Default," and the user's previous selection is lost.
- **Expected:** The user-selected Loop mode should remain unchanged until manually modified again by the user.

## Logs / Screenshots