[SYCL] Align nd_range with SYCL 2020 specifications for next release

[Robertkq]

fixes #22737

[Robertkq]

I believe the __NOEXCEPT macro is a nice way to add noexcept to the specified functions while still being readable.

There are no other uses of __NOEXCEPT as a macro in the project,
if __INTEL_PREVIEW_BREAKING_CHANGES is not defined, the macro expands to nothing

I guess there is a very small chance for downstream users of this header to use __NOEXCEPT, maybe we can call it something more specific to avoid it, like __INTEL_NOEXCEPT or whatever, but I don't think there is a way to not give the macro to downstream users, and I think the readability of this macro outweighs the small chance of collision

EDIT: no longer relevant, ABI Breakage is no concern for this as classes, functions modified are templated, they don't end up in libsycl symbols

[Robertkq]

@KornevNikita can you also take a look? ty

[KornevNikita]

Interesting idea - I'm OK with this, although I'm not sure about runtime team:)

[KornevNikita]

Hey, just checked locally. libsycl.so doesn't contain these symbols, so it seems like we can modify this header without guards

[Robertkq]

Hey, just checked locally. libsycl.so doesn't contain these symbols, so it seems like we can modify this header without guards

Hmm.. I wonder why it doesnt contain these symbols, shouldnt it contain all symbols of the exposed API / Runtime classes? Maybe I dont have enough knowledge on this..

If its convenient for you, can you check symbols for #22736 issue as well?

I'll update the PR accordingly

[KornevNikita]

I see no symbols in https://github.com/intel/llvm/blob/sycl/sycl/test/abi/sycl_symbols_linux.dump and these functions are not marked with __SYCL_EXPORT.

[Robertkq]

@KornevNikita @dm-vodopyanov

Can I have approval for CI & review? simple changes from this PR

Thanks!

[Robertkq]

failiures #22858 #22405