Function reduce_axis 

pub fn reduce_axis<T: Copy + MaybeSendSync, Op: ElementOp<T>, M, R, U>(
    src: &StridedView<'_, T, Op>,
    axis: usize,
    map_fn: M,
    reduce_fn: R,
    init: U,
) -> Result<StridedArray<U>>
where
    M: Fn(T) -> U + MaybeSync,
    R: Fn(U, U) -> U + MaybeSync,
    U: Clone + MaybeSendSync,

Reduce along a single axis, returning a new StridedArray.

Every output folds init once and then its reduced elements. When the reduced axis has unit stride those elements are folded with the lane scheme of reduce, so reduce_fn must be associative and commutative. When the leading kept axis has unit stride, a block of adjacent outputs is accumulated one contiguous source column at a time, which keeps the left to right order per output. With the parallel feature, outputs are split across threads once the number of source elements read exceeds the threading threshold; each output is computed by one thread, so the result does not depend on the thread count.