fn get_best_predictor<T: Pixel>(
fi: &FrameInvariants<T>,
po: PlaneOffset,
org_region: &PlaneRegion<'_, T>,
p_ref: &Plane<T>,
predictors: &[MotionVector],
bit_depth: usize,
pmv: [MotionVector; 2],
lambda: u32,
mvx_min: isize,
mvx_max: isize,
mvy_min: isize,
mvy_max: isize,
w: usize,
h: usize,
) -> MotionSearchResult