[Bug target/126049] RISC-V auto-vectorization misoptimizes fmax()/fmin() regarding NaN

[uwu at icenowy dot me via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=126049

--- Comment #5 from Icenowy Zheng <uwu at icenowy dot me> ---
BTW with Zfa the NaN loading will be optimized to a `fli` instead of loading
from `.word` .