Bug 1543095 and bug 1543098 are implementing events being fired by both Page and Runtime domains, but in an inderteministic order.

Whereas we expect to fire and receive these events in a precise order:

• Runtime.executionContextDestroyed
• Page.frameNavigated
• Runtime.executionContextCreated

The issue is that both Page.navigated and Runtime.executionContextCreated are being fired out of DOMWindowCreated gecko event. The event into which they end up being fired depends on the order into which we call Runtime.enable and Page.enable. These enable methods are registering the DOMWindowCreated listeners.

This work blocks bugs 1543185, 1543151, 1545726 as they are all about using Runtime from another domain.
This work require coordination between Page and Runtime, so I'm expecting this to lay down the typical way two domain should interact between each others.

I submitted a WIP that isn't correct. Instead of coordinating Page and Runtime, it depends on having both domains enabled.
It means that it breaks both domains if one of the two isn't enabled.

I would like to work on this as soon as I'm done with preliminary work related to Runtime, like bug 1548098.

https://treeherder.mozilla.org/#/jobs?repo=try&selectedJob=246150003&revision=fde1e65e9a462da96ef4c518c2f0c420621d92da

Another linux try run after addressing the review comments:
https://treeherder.mozilla.org/#/jobs?repo=try&selectedJob=246150003&revision=0bc183c318a15d209a7870b2ef911b536d85e21d

And a macos run to see if the added test fails on mac:
https://treeherder.mozilla.org/#/jobs?repo=try&selectedJob=246150003&revision=4b2b5f3872ba6e4ec340ea3280184d335c9fecf0
(with a few re-runs)

Note for the sheriffs, if this patch causes any intermittent on Mac, I would prefer to disable the test on this platform rather than backing this out.

https://hg.mozilla.org/mozilla-central/rev/1bb020dacad9