It seems to me that all such occurrences in the ranges-related wording are actually misuses of "ill-formed, no diagnostic required".
In these cases, the phrase denotes runtime requirements for (possibly customized) functions. "Undefined behavior" seems to be a much nicer outcome here.
(Note that "ill-formed, no diagnostic required" could mean a program-wide meltdown as devastating as ODR-violated, which is probably not intended here.)

We chose IFNDR in these cases over UB to allow a sufficiently smart compiler to reject the program if it can prove that a selected operation does not have the required semantics whether or not it can prove that the operation is necessarily executed at runtime.

@CaseyCarter , thanks for the explanation. So, this was intentional and needs an LWG issue to change.

(I disagree with your reasoning, btw; the resulting outcome is much worse than "undefined behavior" and a compiler is free to give warnings or similar even when otherwise successfully compiling the program. Also, the use of IF-NDR for this purpose seems to be rather unique, also among the core language clauses.)

Clarifying: the problem in this issue isn't that IFNDR should be some other behavior, it's that LWG noticed my particular phrasing "ill formed with no diagnostic required" is novel and wanted to harmonize it with other phrasings.

I understand that, but it's still morally the wrong thing to say here.