ref(EX): import upstream ILU kernels + xllm MoE CUDA sources into ex_engine

Copied from upstream_ref (NOT rewritten — exact upstream code):

ixformer C++ API (the authoritative header):
  include/ixformer.h — ixformer::infer namespace: topk_softmax,
    moe_compute_token_index_api, moe_w16a16_group_gemm, moe_expand_input,
    moe_output_reduce_sum, silu_and_mul, rms_norm, xllm_paged_attention, etc.
  include/ilu_ops_api.h — xllm::kernel::ilu namespace: moe_active_topk,
    moe_gen_idx, moe_expand_input, group_gemm, moe_combine_result,
    batch_prefill, batch_decode, rms_norm, matmul, act_and_mul, etc.

ILU kernel wrappers (call ixformer::infer directly):
  csrc/ilu_kernel_fused_moe.cpp — topk routing + gen_idx + expand + combine
  csrc/ilu_kernel_group_gemm.cpp — batched expert GEMM
  csrc/ilu_kernel_{activation,norm,rope,matmul,attention}.cpp

ILU layer implementations (full pipeline):
  csrc/ilu_layer_fused_moe.{cpp,h} — 797 lines, the complete MoE pipeline
    that competitor 168 ran as corex_moe.py
  csrc/ilu_layer_attention.{cpp,h} — prefill/decode attention dispatch

CUDA MoE kernels (from xllm + ds_vllm):
  csrc/moe/moe_topk_softmax_kernels.cuh — CUB BlockReduce + warp topk
  csrc/moe/moe_topk_sigmoid_kernels.cuh — sigmoid scoring variant
  csrc/moe/moe_topk.cuh + moe_fused_topk.cu — entry points
  csrc/moe/moeTopKFuncs.cuh — TRT-LLM derived vllm-compatible topk
  csrc/moe/moe_ops.h + moe_align_sum_kernels.cu — alignment kernels

Common layer headers:
  csrc/common_fused_moe{,_base}.h + common_moe_fused_topk.{cpp,h}

ex_engine/csrc/moe/moe_fused_topk.cu

@@ -0,0 +1,56 @@
+/* Copyright 2026 The xLLM Authors. All Rights Reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    https://github.com/jd-opensource/xllm/blob/main/LICENSE
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+==============================================================================*/
+
+#include "kernels/cuda/cuda_ops_api.h"
+#include "moe_topk_sigmoid_kernels.cuh"
+#include "moe_topk_softmax_kernels.cuh"
+
+namespace xllm::kernel::cuda {
+
+std::tuple<torch::Tensor, torch::Tensor> moe_fused_topk(
+    torch::Tensor& gating_output,
+    int64_t topk,
+    bool renormalize,
+    const std::optional<torch::Tensor>& correction_bias,
+    const std::string& scoring_func) {
+  int64_t num_tokens = gating_output.size(0);
+
+  torch::Tensor topk_weights = torch::empty(
+      {num_tokens, topk},
+      torch::dtype(torch::kFloat32).device(gating_output.device()));
+  torch::Tensor topk_ids =
+      torch::empty({num_tokens, topk},
+                   torch::dtype(torch::kInt32).device(gating_output.device()));
+
+  if (scoring_func == "softmax") {
+    std::optional<torch::Tensor> none_correction_bias = std::nullopt;
+    topk_softmax(topk_weights,
+                 topk_ids,
+                 gating_output,
+                 renormalize,
+                 /*moe_softcapping=*/0.0,
+                 none_correction_bias);
+  } else if (scoring_func == "sigmoid") {
+    topk_sigmoid(
+        topk_weights, topk_ids, gating_output, renormalize, correction_bias);
+  } else {
+    LOG(FATAL) << "Unsupported scoring function for moe topk: " << scoring_func
+               << "only softmax and sigmoid are supported";
+  }
+
+  return std::make_tuple(topk_weights, topk_ids);
+}
+
+}  // namespace xllm::kernel::cuda