[Issue 20198] Make std.math.nextUp and nextDown and nextafter work in CTFE

[d-bugmail at puremagic.com]

https://issues.dlang.org/show_bug.cgi?id=20198

--- Comment #2 from Nathan S. <n8sh.secondary at hotmail.com> ---
For now I will restrict the solution to 64 bit and 32 bit floats. Making
nextUp/nextDown work for extended precision reals in CTFE is difficult because
some D compilers interally use 64 bit precision for CTFE real calculations even
though the target has extended precision reals. This specifically happens in
the FreeBSD_32 and FreeBSD_64_64 tests in the CI suite.

--