Need a concept for closed browsing contexts for bfcache

[annevk]

I initially thought we needed another state only due to the work happening in #3740, but it turns out that child browsing contexts can actually be preserved when navigating. How else could you navigate back with them still being there? (Thanks Nika for pointing this out!)

They mostly need to behave identically to discarded browsing contexts, except that we cannot actually discard them. (I don't have a complete understanding of when the difference matters. It's mostly that they still need to be there.)

(To be clear, this "closed" concept is slightly different from the "is closing" concept, which is solely relevant for close(). And it's reversible in that the browsing context can be "opened" again.)

cc @mystor @bzbarsky

[domenic]

What would this state be used for? I think the fact that the documents are not fully active causes the right consequences for descendants of bfcached pages, usually.

[annevk]

It's not so much about state, it's more that the current model doesn't support this, as far as I can tell. How are they kept around? (See also the other issue where we discussed about the owner of nested browsing contexts.)

[domenic]

Ah, I see. In that case this too is solved by #6315, since "document's browsing context" after the rewrite becomes a strong reference instead of an indirect one.