Entry settings object usage is probably wrong

[domenic]

At least, when I scanned Chrome for usages of the entry settings object, I didn't see anything about XHR in there, and if I recall @bzbarsky did a similar thing for Gecko.

I can try to write up tests for this and submit a patch if my suspicions are confirmed.

[domenic]

Actually, I don't think there's any need to write tests here, as it's impossible for entry/incumbent/relevant/current to mismatch in whether or not they are a Window object. We can just change these all to relevant-of-context-object, or to current, at our leisure. I guess relevant-of-context-object is the way to go.

[annevk]

Current seems nicer? But entry somehow seems purer. It would be a little annoying to change this before the Bikeshed conversion though. Do not want to rebase that across.

[domenic]

Yeah happy to wait until after.

[annevk]

Since you don't want to export entry from HTML, I'm just going to change this to current. I'm not going to use relevant since then I have to change more text and it doesn't matter.

[domenic]

We used relevant for fetch and service worker in this situation, since that's what HTML recommends for consistency.

[bzbarsky]

Fwiw, in terms of implementation Gecko stores the "is an XHR for a window" state during the XHR constructor, by looking at the constructor's current global. But yes, I agree that all of these should be equivalent.