[llvm] Optimize fptrunc(x)>=C1 --> x>=C2 (PR #99475)

[Alexis Engelke via llvm-commits]

aengelke wrote:

This appears to be incorrect w.r.t. round-to-nearest rounding of fptrunc. [alive2](https://alive2.llvm.org/ce/z/w-496p) The constant needs adjustment.

https://github.com/llvm/llvm-project/pull/99475