Removed tentative from compression dictionary tests

[pmeenan]

This removes the "tentative" marker for the compression dictionary transport tests.

The IETF draft has been approved for publication, multiple browsers are working on implementations and is being integrated into the fetch spec: whatwg/fetch#1854

[fred-wang]

@pmeenan

There are merge conflicts and this requires some rebase.

I'm not sure what's the usual time to remove "tentative" but shouldn't we wait that to have a bit more of progress for the spec / tests / implementations?

• The main changes are the IETF draft refers to https://datatracker.ietf.org/doc/html/rfc7234 and fetch PR whatwg/fetch#1854
• whatwg/html#11620 introduces the link relation.
• w3c/webappsec-clear-site-data#94 clarifies clearing cache.

Am I missing any other PRs?

I mentioned on PR 1854 that some tests from dictionary-registration.tentative.https.html are not aligned with the spec.

Also I can't find which spec describes the behavior for "Dictionary registration does not invalidate cache entry" (Vary is only mentioned on https://www.rfc-editor.org/info/rfc9842/#section-6.2 AFAIK):

// Dictionary responses often include
// Vary: available-dictionary, accept-encoding
// We need to make sure that the browser cache does not actually vary
// based on those headers, otherwise a resource that uses itself as a
// dictionary would trigger a second fetch of the same resource.

Also, I suppose the behavior for the "Available-Dictionary header is added when target of redirect matches dictionary" is defined in one of the PR listed above, but that is not immediately obvious to me where. Can you explain it?

(When the spec changes land, I guess it would be nice to add links to relevant spec sections as comments in the WPT tests)