Bugzilla

We should pass https://searchfox.org/mozilla-central/source/testing/web-platform/tests/service-workers/service-worker/ServiceWorkerGlobalScope/unregister.https.html and in fact we do pass it if I run it locally.  I tried to get pernosco to do a self-serve reproduction but unfortunately the pernosco self-serve behavior for Web Platform Tests is weird and it re-runs the whole chunk.

The general test setup is the ServiceWorker [unregister-worker.js](https://searchfox.org/mozilla-central/rev/346ac9335cbc33e3124a4541c938a0dc455e785e/testing/web-platform/tests/service-workers/service-worker/ServiceWorkerGlobalScope/resources/unregister-worker.js) triggers its own unregistration on different lifecycle events and onmessage and the test generally waits for the original `installing` SW to become redundant which should generally happen strictly after we remove the registration (and strictly after we resolve the unregister promise).  Presumably there's something interesting happening with races here and with some local coercing pernosco can get us a good trace.

We should pass https://searchfox.org/mozilla-central/source/testing/web-platform/tests/service-workers/service-worker/ServiceWorkerGlobalScope/unregister.https.html and in fact we do pass it if I run it locally.  I tried to get pernosco to do a self-serve reproduction but unfortunately the pernosco self-serve behavior for Web Platform Tests is weird and it re-runs the whole chunk and fails to reproduce the problem (although sometimes it produces timeouts in other tests in the chunk, which is why pernosco doesn't do that for mochitests).

The general test setup is the ServiceWorker [unregister-worker.js](https://searchfox.org/mozilla-central/rev/346ac9335cbc33e3124a4541c938a0dc455e785e/testing/web-platform/tests/service-workers/service-worker/ServiceWorkerGlobalScope/resources/unregister-worker.js) triggers its own unregistration on different lifecycle events and onmessage and the test generally waits for the original `installing` SW to become redundant which should generally happen strictly after we remove the registration (and strictly after we resolve the unregister promise).  Presumably there's something interesting happening with races here and with some local coercing pernosco can get us a good trace.