refactor: use views NonClientHitTest for draggable regions on mac

[nornagon]

Description of Change

This changes the implementation of draggable regions on macOS to use the
built-in Views tools for it.

See here for some more context.

Checklist

• PR description included and stakeholders cc'd
• npm test passes
• PR release notes describe the change in a way relevant to app developers, and are capitalized, punctuated, and past tense.

Release Notes

Notes: none

[nornagon]

This supersedes #35007, I think.

[zcbenz]

/cc @daniel-koc

[daniel-koc]

Overall looks great.

1. in shell/browser/api/electron_browser_window.cc the following calls seem unnecessary

#if BUILDFLAG(IS_MAC)
  UpdateDraggableRegions(draggable_regions_);
#endif

In such way we could remove overrides for AddBrowserView, RemoveBrowserView, SetTopBrowserView, ResetBrowserViews from BrowserWindow
2. After change:

diff --git a/shell/browser/api/electron_api_browser_window_mac.mm b/shell/browser/api/electron_api_browser_window_mac.mm
index a789647f1a..97c4af3bd3 100644
--- a/shell/browser/api/electron_api_browser_window_mac.mm
+++ b/shell/browser/api/electron_api_browser_window_mac.mm
@@ -19,6 +19,9 @@ namespace electron::api {
 
 void BrowserWindow::UpdateDraggableRegions(
     const std::vector<mojom::DraggableRegionPtr>& regions) {
+  if (window_->has_frame())
+    return;
+
   if (&draggable_regions_ != &regions && web_contents()) {
     draggable_regions_ = mojo::Clone(regions);
   }

The following definitions:

  SkRegion* GetDraggableRegion() const { return draggable_region_.get(); }
  void UpdateDraggableRegions(
      const std::vector<mojom::DraggableRegionPtr>& regions);
  std::unique_ptr<SkRegion> draggable_region_;

we could just move from native_window_mac.h and native_window_views.h into native_window.h

[nornagon]

@samuelmaddock I've dropped one of the tests you added for the focus event here because it doesn't seem to work correctly when using BridgedContentView, which is a part of macOS Views. Based on analysis from @zcbenz, this may be a Chrome bug with the OnWebContentsFocused method, which is fairly lightly used in Chromium itself.

[ckerr]

This looks really good. I have a couple of questions & minor suggestions but both the idea and impl look solid.

[nornagon]

heads up @samuelmaddock, this removes a test that you added because it no longer passes. I'm pretty sure this is due to a Chrome bug that doesn't properly emit the web contents focused event when the window is focused on Mac/Views—the event is emitted if I click on the WebContents, just not when the window is only focused. Nothing else seems to be problematic in terms of using the WebContents though.

[release-clerk]

No Release Notes