CUDA optimization speeds up prompt processing to 2600 t/s.

The official source reports this update: b10704: CUDA: use the fast mm_ids_helper path for any n_expert_used. CUDA: use the fast mm_ids_helper path for any n_expert_used ( #27978 ) The optimized path grouped warp lanes by token and required warp_size % n_expert_used == 0, with a single hardcoded exception padding 6 up to 8. Every other count fell back to the generic path, which walks the tokens one at a time with a warp reduction per token, for each of the n_expert blocks. The lane group only has to divide the warp, and the loop body already guards the padded lanes with iex generalizes to the next power of two.