unsafe_copy3d! requires 2^4 alignment

[jpsamaroo]

As reported by HSA in CI on master. We should probably switch to a kernel copy for the portions which aren't aligned.

[luraess]

Is this still the case? (cc @jpsamaroo @pxl-th) Maybe worth checking as I saw the function got refactored in #374

[pxl-th]

hsa_amd_memory_async_copy_rect accepts hsa_pitched_ptr_s docs for which still say that alignment is required.

We may however, add hsa_amd_memory_async_copy for regular async copy.