[clang] [llvm] [AMDGPU] Add async variants of tensor load/store LDS intrinsics. (PR #200775)

[Pierre van Houtryve via cfe-commits]

Pierre-vh wrote:

> I am strongly opposed to any use of that legacy convention on that.

I strongly agree too. We should do what's best for the compiler and UX, not follow some arbitrary convention from the past.
If there is a strong appetite for "direct" intrinsics that map 1-1 to ISA, they should exist in a separate namespace like "amdgpu.footguns".

https://github.com/llvm/llvm-project/pull/200775