Re the behavior of std::strong_order for floating-point types, see
https://stackoverflow.com/questions/59348310/how-to-implement-the-totalorder-predicate-for-floating-point-values
https://stackoverflow.com/questions/69068075/what-is-the-purpose-of-if-x-and-y-represent-the-same-floating-point-datum-i
https://quuxplusone.github.io/blog/2021/09/05/float-format/
There's still an open question of what it should do for long double; I'm proposing to just leave it unimplemented in hopes that someone who cares about it will provide an implementation.
This PR does not implement std::strong_order_fallback et al. That will be a second PR. (It is now D111514.)
Do you mean 14.0 or 13.0 here?