[llvm] [Offload] Add olLaunchKernelSuggestedGroupSize (PR #142130)

[Callum Fare via llvm-commits]

callumfare wrote:

I'd lean towards only providing the information needed to make the decision for the preferred sizing at the liboffload level - something like a device query for `cuOccupancyMaxPotentialBlockSize` and/or whatever else is helpful. The final recommended size might depend on language related factors that liboffload doesn't know about (reqd_work_group_size in SYCL, the Kernel Environment in OpenMP) so I think it's better to leave it to the next level up the stack. And if we do that then a special olLaunchKernel variant doesn't really make sense.

https://github.com/llvm/llvm-project/pull/142130