P3567R2 flat_meow fixes

[eisenwave]

Fixes #8463

No NB comment.

Also fixes cplusplus/papers#2211

PROBLEM: cplusplus/papers#2029 added constexpr to the members of flat_meow, but what we're merging doesn't have any constexpr on it. LWG issue? @jwakely

[jwakely]

There's no need for an issue, the intended design is clear: they should all be constexpr. Please just add constexpr keywords as a fixup commit.

[jensmaurer]

Please just add constexpr keywords as a fixup commit.

Yes, please.

[eisenwave]

Alright, I've made them all constexpr in a new commit. It's not a fixup commit because I think it's worth having some record of why we deviate from the paper in the commit history, but Thomas can squash it anyway if he doesn't think so.

[jwakely]

Suggested change

	$N$ + $M \log M$, where $N$ is \tcode{size()} before the operation and $M$
	is \tcode{ranges::distance(rg)}.
	$N$ + $M \log M$, where $N$ is \tcode{size()} before the operation
	and $M$ is \tcode{ranges::distance(rg)}.

(semantic break)

[eisenwave]

The chosen line break is consistent with what we have further up for flat_map::insert and further below for flat_set::insert.

The and here is also a conjunction in a list, so it seems more appropriate to put it at the end of the line.

[eisenwave]

Should be all fixed now.

Had to force-push because the second commit which adds constexpr would be hard to merge with fixups that change sorted_unique_t to sorted_equivalent_ otherwise.