…into main

* 'main' of https://github.com/sgl-project/sgl-kernel-npu: (44 commits)
  fix a2 deepep doc (sgl-project#279)
  prepare build and release for a2 (sgl-project#273)
  add deepep a2 doc (sgl-project#277)
  Add the long-sequence ant migration feature for the prefill combine operator. (sgl-project#267)
  Fixing Chinese character encoding issues (sgl-project#275)
  [Bugfix] fix TorchNpuHelper rename bugs (sgl-project#265)
  qwen3-next op optimize (sgl-project#257)
  fixup bug in conv1d_update_fn (sgl-project#259)
  add long sequence feature for normal deep_ep (sgl-project#254)
  modify md file (sgl-project#255)
  sgl-kernel-npu add release version (sgl-project#253)
  add a script for generalize test (sgl-project#131)
  fixing release ci (sgl-project#248)
  fix normal and low_latency layerd rdma_data_size when mixed running (sgl-project#246)
  Fixing the issue where the A2 notify_dispatch operator gets stuck on cann8.3 (sgl-project#245)
  rework release build (sgl-project#237)
  release build (sgl-project#231)
  Add two mixed-race tests: normal and low latency, normal and fused deep moe. (sgl-project#206)
  [Feat] add chunk_gated_delta_rule triton support (sgl-project#232)
  [Bugfix] add padding cases for causal_conv1d_update (sgl-project#235)
  ...