@jwakely, @cor3ntin: Could you please comment on the Editor's Note in the paper that asks whether we need to modify the container requirements, and says,

we probably want to requiere is_constructible<value_type, T> where T is either the type passed to insert, or
the InputIterator’s value_type. Currently, we only seem to require convertible_to, which may
not be sufficient?. An alternative is to add explicit insert overloads for pair-like objects

@tkoeppe Sorry, I guess i should have removed that note, it was more "lwg should discuss that" rather than actual editor guidance. I don't actually remember that it was discussed though.

I don't think there is an issue here either way, I'm struggling to imagine a case where pair would not be constructible from tuple-like. We may still want to add new overloads to associative containers for performance, maybe (i need to think about it more) but this would be nb comment material.

@cor3ntin: Thank you! I'll proceed as-is then, and we can follow up via issues later if and when necessary.

The 2nd part of this (replacing all usages of tuple-or-pair by tuple in the range clause) should be applied after P2374R4 #5605 as per the Editor's note at end of P2165R4.

I will add a corresponding addition to that motion when we get to it; for now, this paper is applied in its entirety.

Also resolves cplusplus/papers#876.