Potential Windows-specific lost event race in the implementation of time.sleep

Bug description:

In the Windows-specific implementation of time.sleep there appears this code:

https://github.com/python/cpython/blob/b706ff003c536c5bca24dfdd3a8917bffcfa3df1/Modules/timemodule.c#L2341-L2345

If a control-C event happens after PyErr_CheckSignals returns, but before ResetEvent is called, I believe that event may be lost.

This is probably a hard race window to hit, and I don't have a Windows development environment, so I have not confirmed that it is possible.

CPython versions tested on:

CPython main branch

Operating systems tested on:

Windows

先從連結行中的 Modules/timemodule.c 裡 Windows 專用的 time.sleep 實作開始，然後檢查 PyErr_CheckSignals 和 ResetEvent 的執行順序。調查在 Windows 上 control-C 事件是否可能在該時間窗口內發生；確認或排除遺失事件的競態，並在適用時透過覆蓋測試記錄或修正該行為，即視為完成。