[llvm] [AMDGPU] Add intrinsic for raw atomic buffer loads (PR #97707)

[Jessica Del via llvm-commits]

OutOfCache wrote:

> Two tests added in this commit seem to be failing on a build bot [lab.llvm.org/buildbot/#/builders/174/builds/2102](https://lab.llvm.org/buildbot/#/builders/174/builds/2102). Can you take a look and revert if you need time to investigate?

> The same thing is happening on AArch64 buildbots: [lab.llvm.org/buildbot/#/builders/125/builds/867](https://lab.llvm.org/buildbot/#/builders/125/builds/867)

Thanks for letting me know. Seems like mostly the register numbers changed in the check lines. Here is a PR for the fix: https://github.com/llvm/llvm-project/pull/99912

https://github.com/llvm/llvm-project/pull/97707