Function widen

pub(in ::num::bfloat) const fn widen(value: f16b) -> f32