Drop "keyevents" as an alias of "keyboardevent" for createEvent

[foolip]

There's a test for this in wpt and a proposed fix in Blink:
http://w3c-test.org/dom/nodes/Document-createEvent.html
https://codereview.chromium.org/1569813002/

As discussed in that review, it looks like this alias is only supported by Gecko, and so we think it would be better to drop it from the spec. If it turns out to be unsafe to remove in Gecko, then the only way out is for all other engines to add it.

[dtapuska]

TextEvent already behaves similar too. Gecko supports TextEvents; but nobody else does. Of course TextEvent isn't part of the current spec and is deprecated; but nonetheless there are other examples of things Gecko supports that Safari, Edge, and Blink based browsers don't support.

[annevk]

Paging @smaug---- for thoughts.

[foolip]

TextEvent isn't in http://w3c-test.org/dom/nodes/Document-createEvent.html, fortunately.

Blink/WebKit are also very naughty, by supporting all known event interface names for createEvent, not just the ones listed in the spec.

[smaug----]

Gecko used to have only keyevents, and keyboardevent was added in
https://bugzilla.mozilla.org/show_bug.cgi?id=238773, so, long ago.
In mozilla-central there are several tests using "keyevents", so I guess would need to fix those and then add some telemetry and based on that say something here.

[foolip]

How about removing it from the spec and adding it back only if there's evidence that it's needed for web compat, e.g. if Gecko fails to remove it? Leaving it in risks that someone implements it, if a reviewer doesn't pay close attention.

[annevk]

That sounds reasonable to me. @smaug---- is there a bug against Gecko on removing this?

[ArkadiuszMichalski]

I open bug for Gecko:
https://bugzilla.mozilla.org/show_bug.cgi?id=1261874