SYCL_KHR_QUEUE_FLUSH implementation and optional extension mechanism

[PeterTh]

This is a continuation of #44, adding the ability to configure the availability of the extension, as well as (and this is the larger part) a new CI action which checks that this mechanism is actually working.

[PeterTh]

@TApplencourt please have a quick look to see if this suits your needs, if so I'll merge it.

[TApplencourt]

Thanks a lot for your time on it!

LGTM, except that I don't really understand why the CI doesn't have a Configure CMake for extension check program (with extensions) , but my CMake skills are poor, and I didn't even get my coffee : )

Please merge it at your convenience, or you can also wait for the SYCL WG to merge the KHR first if you prefer to be "save",

Thanks again

[PeterTh]

LGTM, except that I don't really understand why the CI doesn't have a Configure CMake for extension check program (with extensions)

That's a valid question, because the lack of symmetry is jarring -- I actually considered doing that step for that reason despite it not being required and not actually doing anything.

The reason it is not necessary is because the configuration change will change the installed SimSYCL the test is built against, but it's still in the same location. So the CMake configuration of the test does not need to change.

I think I'll wait with merging until the KHR is merged.