[llvm] [AMDGPU] Make maximum hard clause size a subtarget feature (PR #81287)

[Krzysztof Drewniak via llvm-commits]

krzysz00 wrote:

@jayfoad Yeah, the line between what goes in the `.td` and what gets a hardcoded function in `GCNSubtarget.h` is unclear

Should we go ahead and land this?

https://github.com/llvm/llvm-project/pull/81287