[llvm] [RFC][AMDGPU] Remove old llvm.amdgcn.buffer.* and tbuffer intrinsics (PR #93801)

[Jay Foad via llvm-commits]

jayfoad wrote:

@llvm/pr-subscribers-mlir-amdgpu I see the MLIR build failed with:
```
mlir/include/mlir/Dialect/LLVMIR/ROCDLConversions.inc:142:28: error: no member named 'amdgcn_buffer_load' in namespace 'llvm::Intrinsic'; did you mean 'amdgcn_s_buffer_load'?
mlir/include/mlir/Dialect/LLVMIR/ROCDLConversions.inc:151:28: error: no member named 'amdgcn_buffer_store' in namespace 'llvm::Intrinsic'; did you mean 'amdgcn_raw_buffer_store'?
```
Does MLIR really depend on these old intrinsics? Can it be fixed?

https://github.com/llvm/llvm-project/pull/93801