Add a few tests for whether CSS subresources are critical

[gsnedders]

tests for w3c/csswg-drafts#1088

these should be moved somewhere better once we know what directory to put them in

[gsnedders]

While the CSS WG has still not resolved the above issue, these tests do at this point show identical behaviour across all three major browser engines, hence on the whole I'd be in favour of landing these as tentative tests if someone wants to review them.

[domfarolino]

For some reason this thread was buried in my inbox and it doesn't look there's been much activity in a while. Are you still seeking review @gsnedders? Has there been any more consensus or spec updates here?

[gsnedders]

Has there been any more consensus or spec updates here?

Not that I'm aware of; that said, I'm still in favour of review (and landing) the tests which pass in all three major browser engines, as we may as well have tentative tests that assert the current intersection behaviour (and ensure we don't move away from interoperability there).

[domfarolino]

Generally LGTM % a couple of comments. Also, maybe we can change instances of var to let or const appropriately?

[domfarolino]

Is this <script></script> needed?

[domfarolino]

In 2/4 font-using tests in this PR, the id=foo element comes before the <style> that includes the font, and in the other 2/4 it is the opposite. Can we make this consistent? (Not that it really matters much...)

[domfarolino]

Another variation for these tests that could be useful to include in this PR is: moving all of the <style> block contents to a <link rel=stylesheet> and seeing if the link element's load event is indeed blocked when the Document's load event is blocked.

In w3c/csswg-drafts#1088 (comment), Hiroshige points out that apparently Chrome blocks the Document's load event in some cases where the <link> element's load event is not blocked, which is pretty interesting. It would be good to know what browsers do here more generally.

[gsnedders]

Another variation for these tests that could be useful to include in this PR is: moving all of the <style> block contents to a <link rel=stylesheet> and seeing if the link element's load event is indeed blocked when the Document's load event is blocked.

I'll fix the nits you mentioned above, but I probably won't expand the coverage of this (I don't really immediately have the time to actually work on this right now).

[domfarolino]

Let me know whenever you do and I'm happy to re-review