Fold `min/max(fpext x, C)` to `fpext(min/max(x, fptrunc C))` in cases where the truncation of the constant is lossless. This helps eliminate fpext/fptrunc pairs around min/max and addresses the regression from https://github.com/llvm/llvm-project/pull/177988. Proof: https://alive2.llvm.org/ce/z/y_Bcdd