iOS: Never queue application-level events

[madsmtm]

Events like resumed, new_events, about_to_wait, and so on will never happen as a result of programmer action, so we'll never need to queue those. This allows us to remove the need for the old Event struct in the iOS backend.

Furthermore, we can now remove InUserCallback, since that state is already stored inside EventHandler.

I've tried to otherwise keep the semantics as close to the original by calling handle_nonuser_events(mtm, []), which flushes pending events.

• Tested on all platforms changed

[madsmtm]

Incidentally, this improves #3714, but makes #3816 much worse - not quite sure why yet? But will fix it separately from this, see #3906

[kchibisov]

Commit should specify that it fixes specific issue you're talking about and you should have a changelog entry for it.

[madsmtm]

I would treat the fact that it fixes #3714 as more of a curiosity, not a true fix, there's still a bunch of things I need to do to fully fix that issue. But I have added a changelog entry.

[kchibisov]

I would treat the fact that it fixes #3714 as more of a curiosity, not a true fix, there's still a bunch of things I need to do to fully fix that issue. But I have added a changelog entry.

In this case it shouldn't matter and it's better without it. And saying that it fixes it should be removed, since it's not as you say.

[madsmtm]

Sure, I've updated the comment

[kchibisov]

Does the changelog entry really matter at this point? Maybe just drop it, since it's an implementation detail. I thought that you've fixed the issue meaning that you'd close it with a changelog entry, but if it's not fixed, then changelog entry doesn't really matter.

[madsmtm]

That's why I didn't have a changelog entry in the first place ;)

But I think there is value in keeping the entry, while this is an implementation detail, it can be useful for users to know if something were to go wrong with it.

[kchibisov]

I just don't see how it's actionable or solves issues, and I've asked for changelog only because you've said that it fixes some old bug. I'll leave it up to you though.

[madsmtm]

Well, at the very least it should improve performance (once #3816 is also resolved), just because we do less. So just for that reason, the changelog entry is nice.