[llvm] [InstCombine] Canonicalize reassoc contract fmuladd to fmul + fadd (PR #90434)

[Matt Arsenault via llvm-commits]

arsenm wrote:

> hi @andykaylor，does it have some conclution on this topic at last FP working group meeting? Is the reassoc flag is necessary?

I think we were leaning towards removing the combine that's already here, not extending it. We should treat fmuladd like an fma intrinsic, and any regressions need to be addressed by dealing with fma/fmuladd in other contexts 

https://github.com/llvm/llvm-project/pull/90434