AMDGPU: Replace sqrt OpenCL libcalls with llvm.sqrt (#74197)

The library implementation is just a wrapper around a call to the
intrinsic, but loses metadata. Swap out the call site to the intrinsic
so that the lowering can see the !fpmath metadata and fast math flags.

Since d56e0d07, clang started placing
!fpmath on OpenCL library sqrt calls. Also don't bother emitting
native_sqrt anymore, it's just another wrapper around llvm.sqrt.