Correct assorted issues with standalone pseudo elements

[annevk]

I discovered these issues while writing WebKit/WebKit#22231. Standalone pseudo-elements still need to be properly formatted.

And ::placeholder is a valid pseudo-element.

Closes w3c/csswg-drafts#9751.

[annevk]

The negative case of the ::view-transition stuff is covered by https://github.com/web-platform-tests/wpt/blob/master/css/cssom/getComputedStyle-pseudo.html. In particular the :marker test. Could probably have more coverage though.

[nt1m]

We should have a test that this is invalid, I know @karlcow had created a test about getComputedStyle + pseudo elements, so it could be added there.

[karlcow]

wpt/css/cssom/getComputedStyle-pseudo.html

Lines 161 to 175 in ee42dc1

	test(function() {
	let li = document.querySelector('li');
	assert_true(
	getComputedStyle(li, ':marker').length == 0,
	"Should return an empty style for :marker");
	assert_true(
	getComputedStyle(li, 'marker').length != 0,
	"Should return the element style for marker");
	assert_equals(
	getComputedStyle(li, 'marker').color, "rgb(255, 0, 0)",
	"Should return the element style for marker (ex: color is rgb(255, 0, 0), not rgb(0, 128, 0)");
	assert_equals(
	getComputedStyle(li, '::marker').color, "rgb(0, 128, 0)",
	"Should return the element style for marker (ex: color is rgb(0, 128, 0)");
	}, "Unknown pseudo-element with a known string (ex: marker)");

[annevk]

Yeah, I also pointed to this above. I think that's okay for now, but we could add more coverage.

[nt1m]

We could change that test to loop over different pseudo elements

[karlcow]

we could probably cover all of them. That would make sure nothing bad is happening. With the loop being outside of the test. So we know which pseudo-element fails instead of failing everything if one is missing.

[nt1m]

Not sure why this was considered invalid

[annevk]

I think they previously prohibited -webkit-prefixed pseudo-elements and maybe this is a leftover of that, but not sure.