Dan Hoeflinger

I'm going to hold on this PR until we have a system solid within the CI to handle it, in case other changes are found to be needed.

> @danhoeflinger, have you also considered replacing/complementing CMake checks of a compiler? It may be a more nature way to follow, compared to adding a new test along with other...

update on this... It seems that inclusion of a broken tbb install may be the bigger issue in our nightly windows CI testing. This may still be a good idea...

I'll further note a compounding issue that for both of these cases linked above, we have a runtime decision to dispatch to 1 of 2 kernels, where in each case...

I realize I never expanded the set of reviewers from a small group, though I know we agreed not to attempt to spend too much effort on perfecting this approach...

Transform_if uses `read_write` here to force a copy-in currently, even though read access isn't _really_ required. What it wants is `write` without `no_init`. This becomes more of a problem in...

> I have a memory of us running into issues with some compilers providing `sycl.hpp`, but not actually providing a working SYCL implementation when we were adding this to CMake...

I'm investigating the failure for windows, cl, but I'm thus far unable to reproduce it...

> I'm investigating the failure for windows, cl, but I'm thus far unable to reproduce it... I'm still unable to reproduce the failure in practice. However, examining the code a...

> I'm still unable to reproduce the failure in practice. However, examining the code a bit more, I think that our usage of `::new (::std::addressof(*__result)) _Tp(*(__select_element(__first1, __first2)));` is generally incorrect...