[sgl-kernel] Add cuda kernel for moe_ep_silu_and_mul (#6919)

Co-authored-by: luoyuan.luo <luoyuan.luo@antgroup.com>

sgl-kernel/python/sgl_kernel/__init__.py

@@ -52,6 +52,7 @@ from sgl_kernel.moe import (
     cutlass_fp4_group_mm,
     ep_moe_post_reorder,
     ep_moe_pre_reorder,
+    ep_moe_silu_and_mul,
     fp8_blockwise_scaled_grouped_mm,
     moe_align_block_size,
     moe_fused_gate,

sgl-kernel/python/sgl_kernel/moe.py

@@ -88,6 +88,24 @@ def ep_moe_pre_reorder(
     )
 
 
+def ep_moe_silu_and_mul(
+    gateup_output,
+    down_input,
+    reorder_topk_ids,
+    scales,
+    start_expert_id,
+    end_expert_id,
+):
+    return torch.ops.sgl_kernel.ep_moe_silu_and_mul.default(
+        gateup_output,
+        down_input,
+        reorder_topk_ids,
+        scales,
+        start_expert_id,
+        end_expert_id,
+    )
+
+
 def ep_moe_post_reorder(
     down_output,
     output,