[WIP] feat: deep research demo

[defi-failure]

What this PR does

这个pr参照 https://github.com/langchain-ai/open_deep_research 初步实现了一个 multi-agent deep research demo

基本结构如图

当前效果如图（长图已折叠）

目标：让所有支持function call的模型都拥有deep research能力
当前状态：流程已初步跑通，当前测试过效果最好的是gemini 2.5 pro/flash+思考模式+brave search tool

已知问题与后续工作：

1. 错误处理还存在一些问题，正在运行的researcher无法立即停止；没有考虑rate limit，免费的gemini key可能会在某个点触发too many requests
2. prompt尤其是clarify with user和generate reserch brief的prompt需要再进行一定的调整，当前老是会出现在generate reserch brief阶段要求用户澄清，并且生成的research brief也往往不全面
3. 多个researcher无法真正并行执行任务
4. ui显示比较乱，开始研究前没有确认步骤，中间结果和思考过程大都显示了，会显得比较乱，另外最终显示报告的组件也没有像gemini，openai或者kimi那样是定制组件
5. token计算暂时只会显示最后一条消息的
6. 关于搜索目前只测试过gemini的内置搜索，cherry自带的搜索支持，和brave search mcp tool，里面表现最好的是brave search mcp tool，gemini内置搜索因为gemini的grounding机制引文不准确的现象会很严重
7. i18n还没有做
8. researcher目前只能利用搜索和思考工具，如果可以像kimi deep research那样扩展到python code，并且可以生成交互式的报告应该会很棒

Fixes #

Why we need it and why it was done in this way

The following tradeoffs were made:

The following alternatives were considered:

Links to places where the discussion took place:

Breaking changes

If this PR introduces breaking changes, please describe the changes and the impact on users.

Special notes for your reviewer

Checklist

This checklist is not enforcing, but it's a reminder of items that could be relevant to every PR.
Approvers are expected to review this list.

• PR: The PR description is expressive enough and will help future contributors
• Code: Write code that humans can understand and Keep it simple
• Refactor: You have left the code cleaner than you found it (Boy Scout Rule)
• Upgrade: Impact of this change on upgrade flows was considered and addressed if required
• Documentation: A user-guide update was considered and is present (link) or not required. You want a user-guide update if it's a user facing feature.

Release note

[eeee0717]

Pretty Good!

[EurFelux]

Good Work!

[beyondkmp]

Note

This comment was translated by Claude.

Awesome!

---

Original Content

牛逼