fix(windows): dangling pointer when remove_for_hwnd called before drop

[Legend-Master]

Fix tauri-apps/tauri#15523

Spent so much time tracking this one down, the problem in tauri-apps/tauri#15523 was not that the menu getting accessed off main thread but simply the drop orders

A simpler reproduction tauri code is just

{
    let _ = window.remove_menu().unwrap();
}
{
    let _ = window
        .set_menu(tauri::menu::Menu::default(&app_handle).unwrap())
        .unwrap();
}

The problem here is that we didn't clean up the window subclass when removing the menu (the drop didn't clean it up because the tracking hwnd is removed by remove_for_hwnd) and also we didn't refresh it in time that we called SetMenu before SetWindowSubclass in init_for_hwnd_with_theme, and SetMenu triggered menu_subclass_proc which contains the stale Menu pointer

---

sorry about the push to dev branch by accident

[github-actions]

Package Changes Through de615bc

There are 1 changes which include muda with patch

Planned Package Versions

The following package releases are the planned based on the context of changes in this pull request.

package	current	next
muda	0.19.2	0.19.3

Add another change file through the GitHub UI by following this link.

---

Read about change files or the docs at github.com/jbolda/covector

[Legend-Master]

I swear I wrote the change file and it didn't look like this 🤦‍♂️