[SYCL][Bindless][Exp] Fix get_mip_level_mem_handle declaration.

[tangjj11]

A declaration of get_mip_level_mem_handle incorrectly took sycl::device as a parameter, which should have been a sycl::queue. This is now fixed.

[isaacault]

Is this not an ABI break? I would've expected CI to catch that and fail..

I'd think that the right way to do this would be to deprecate this definition and export the new, correct, definition, although with CI passing I'm not sure what to think. Perhaps we should deprecate anyway, to be proper? Or am I misunderstanding?

[tangjj11]

Is this not an ABI break? I would've expected CI to catch that and fail..

I'd think that the right way to do this would be to deprecate this definition and export the new, correct, definition, although with CI passing I'm not sure what to think. Perhaps we should deprecate anyway, to be proper? Or am I misunderstanding?

I thinks that the reason why the CI has not catch the ABI breaking fail is the same as @przemektmalon's reply in #11948.

[tangjj11]

@intel/llvm-gatekeepers can this be merged? thanks

[sommerlukas]

Unrelated test failures:

• Unexpected passing test USM/usm_leak_check.cpp
• Failure in basic_tests/SYCL-2020-spec-constants.cpp
• sycl-ls failure on Windows