[Hardware][NVIDIA] FP4 MoE kernel optimization (#19110)

Signed-off-by: Chiyue Wei <chiyuew@nvidia.com>
Co-authored-by: Chiyue Wei <chiyuew@nvidia.com>

csrc/torch_bindings.cpp

@@ -450,7 +450,7 @@ TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, ops) {
       "                        Tensor! problem_sizes1, Tensor! problem_sizes2, "
       "                        Tensor! input_permutation, "
       "                        Tensor! output_permutation, int num_experts, "
-      "                        int n, int k) -> ()",
+      "                        int n, int k, Tensor? blockscale_offsets) -> ()",
       {stride_tag});
   ops.impl("get_cutlass_moe_mm_data", torch::kCUDA, &get_cutlass_moe_mm_data);