[llvm] [LLVM][AArch64][tblgen]: Match clamp pattern (PR #75529)

[Paul Walker via llvm-commits]

https://github.com/paulwalker-arm commented:

Sorry for the back and forth but I think this lastest version of the patch has taken a slightly backward step.

LLVM has both sorts of floating-point max/min operations (see https://www.llvm.org/docs/LangRef.html#llvm-maxnum-intrinsic) so you should be able to keep your previous PatFrags but swap out AArch64fmax_p for AArch64fmaxnm_p by using the `llvm.maxnum` intrinsic in your original tests.

https://github.com/llvm/llvm-project/pull/75529