[llvm] [AMDGPU] Define constrained multi-dword scalar load instructions. (PR #96161)

[Christudasan Devadasan via llvm-commits]

cdevadas wrote:

> I also must say that s_buffer_load is in the same bucket.

Buffer loads are generated for the buffer intrinsic and we take care of them during legalization to always use natural alignment. But we might require the *_ec buffer opcodes for SILoadStoreOptimizer while merging them. @rampitec do you think that can happen?

https://github.com/llvm/llvm-project/pull/96161