some small langref changes

[IOKG04]

1. Stating that sentinel-terminated slices can be coerced into sentinel-terminated pointers. (closes Still more slice/pointer confusion #9628)
2. Currently it is stated that error return traces are enabled by default in ReleaseSafe. This is false and now corrected. (closes @errorReturnTrace() == null in ReleaseSafe build #24232)
3. Currently the langrefs for @rem() and @mod() state that the caller must guarantee denominator > 0, when it is actually denominator != 0, which is what it's changed to. (closes @rem and @mod langref documentation is wrong #23635)

Regarding 3. I also wanted to ensure the new langrefs would stay correct, so I looked if denominator < 0 is tested anywhere. For @mod() it is already tested, for @rem() not, so I added those tests.
I hope I added them in the correct file, if not I can change it.

[sfiedler0]

I guess it would also be a great idea fix a mistake in @export where the test says .linkage = .strong, but the explanation says "options.linkage is '.Strong'" (so in the documentation there is a case error).
(and yes, I could have opened a separate pull request and issue, but maybe you could also do it here so there is just one pull request)

[leecannon]

setFloatMode also refers to Strict and Optimized, both of these should be lowercase

[IOKG04]

setFloatMode also refers to Strict and Optimized, both of these should be lowercase

In that case, wouldn't the floating point operator documentation also have to use lowercase versions?
Also I feel like the uppercase is understandable there, considering it isn't prefixed with ., meaning it could just be referring to the name, not the code.

[mlugg]

LGTM, thanks!