Feature/any all

[stevenrbrandt]

This PR allows the code despite a compiler bug in cuda 12.9.1 https://bitbucket.org/einsteintoolkit/tickets/issues/2922/carpetx-does-not-compile-with-cuda-129

It moves any/all to namespaces and fully qualifies their use in a few cases.

[rhaas80]

LGTM

[eschnett]

I believe my recent PR that removes using namespace std fixes this. The problem is that C++23 defines a class std::any (and std::all), and this class is accidentally in scope because some header files use using namespace std. This is also the reason why things are failing only for any and all, and not for any of the other functions.

[rhaas80]

I see. So the nvcc compiler on the failing cluster defaulted (or faultily used) to C++23 (since we should be asking for -std=c++17?

[eschnett]

I think the compiler is using C++17, but the standard library is using C++23, probably simply because you're using a newer version of GCC (or nvcc).

[rhaas80]

I believe my recent PR that removes using namespace std fixes this. The problem is that C++23 defines a class std::any (and std::all), and this class is accidentally in scope because some header files use using namespace std. This is also the reason why things are failing only for any and all, and not for any of the other functions.

#381 is the PR? That pull request unfortunately has build failures (missing std:: in setprecision call in some io thorn. I will try and update it.

[rhaas80]

Superseded by #381

[rhaas80]

Superseded by #381