Add a feature that automatically waits for a model provider's rate limit to e...

Add a feature that automatically waits for a model provider's rate limit to expire and retries the request instead of immediately failing.

---
| Field | Value |
| --- | --- |
| App version | 1.0.25 |
| OS | Windows 10.0.26200 |
| Theme | ABOLKOG |
| Path | /chat |
| Tenure | Week 2 |

首先跟踪 /chat 请求路径，并定位 model-provider rate-limit failure 的处理位置。确定 provider 如何传达重试延迟，以及重试应如何影响用户可见的请求流程；完成标准是：被 rate limit 的请求会等待并重试，而不是立即失败，同时在其他情况下保留现有的失败行为。

由索引模型根据 Issue 内容生成。