Improved headless Servo performance

[mateon1]

Now the main thread doesn't waste 100% CPU, polling for events.

---

• ./mach build -d does not report any errors
• ./mach test-tidy does not report any errors
• These changes fix #18770 (github issue number if applicable).

• There are tests for these changes OR
• These changes do not require tests because: I verified Servo works correctly in headless mode, and the linked issue is gone.

---

This change is 

[highfive]

Heads up! This PR modifies the following files:

• @paulrouget: ports/glutin/window.rs

[paulrouget]

This is good enough for tests. If headless mode is only used for tests, I think this approach is ok.

But if Headless mode is used for anything else, we need a better approach. I think the proper way of doing this is:

• In Window::wait_event , for Headless, where non-headless do: window.poll_events(), we should block (park the main thread)
• in create_window_proxy, for Headless, implement a thread-safe object that will wake up the blocked thread

Then the loop would run only if needed.

[nox]

@bors-servo try

[bors-servo]

⌛ Trying commit 35b7527 with merge 40014c1...

[paulrouget]

@jdm do you think we should just add this sleep call, or do as I describe in my previous comment (less trivial)?

I'm happy with both approaches.

[jdm]

Why don't we make this change right now and file an issue for doing it the proper way?

[paulrouget]

Ok.

@mateon1: code looks good. Maybe move the sleep in WindowKind::Headless(..) => {} and add a comment.

[paulrouget]

Thank you.

I filed #18799

Can someone r+ this?

[jdm]

@bors-servo r+

[bors-servo]

📌 Commit 63d69db has been approved by jdm

[bors-servo]

⌛ Testing commit 63d69db with merge 54f6e87...

[bors-servo]

☀️ Test successful - android, arm32, arm64, linux-dev, linux-rel-css, linux-rel-wpt, mac-dev-unit, mac-rel-css1, mac-rel-css2, mac-rel-wpt1, mac-rel-wpt2, mac-rel-wpt3, mac-rel-wpt4, windows-msvc-dev
Approved by: jdm
Pushing 54f6e87 to master...