[QST] Result layout of logical_product is ((M,N), (TileM,TileN), L, ...)?

From docs: https://github.com/NVIDIA/cutlass/blob/main/media/docs/cute/02_layout_algebra.md, the result layout of logical_product is ((M,TileM), (N,TileN), L, ...). But from some example and code, i think the layout maybe is ((M,N), (TileM,TileN), L, ...). Do i miss something?

Code:
https://github.com/NVIDIA/cutlass/blob/main/include/cute/layout.hpp#L1662
make_layout(block, composition(complement(block, size(block)*cosize(tiler)), tiler)); // block layout is at first place.

example1 :
auto thr_layout = make_layout(make_shape(Int<32>{}, Int<8>{})); auto val_layout = make_layout(make_shape(Int<4>{}, Int<1>{})); auto layout_mn = logical_product(thr_layout, val_layout); --> result: ((_32,_8),(_4,_1)):((_1,_32),(_256,_0)) // block is at first.

example2 :
auto vec = Layout<Shape<_2, _5>,Stride<_5, _1>>{}; auto tile = Layout<Shape<_3, _4>,Stride<_1, _3>>{}; auto layoutR = logical_product(layoutA, layoutB); --> result: ((_2,_5),(_3,_4)):((_5,_1),(_10,_30)) // block is at first.

Start with media/docs/cute/02_layout_algebra.md and include/cute/layout.hpp around line 1662. Reproduce the two logical_product examples and compare their reported shapes with the documentation and implementation. Done means explaining the ordering consistently and updating the documentation or examples if they disagree.