ref(upstream): FULL TREE — Deep-Spark xllm (1470) + ds_vllm csrc/models (703)

Replaces cherry-picked upstream_ref with complete source trees.

xllm/ — Iluvatar official C++ inference engine (15MB, 1470 files)
  Complete: kernels → layers → models → runtime → scheduler → api
  Excluded: .git, binary images, third_party submodule checkouts

ds_vllm/ — Iluvatar official vllm fork (8MB, 703 files)
  Included: csrc/ (ALL CUDA kernels), fused_moe/, qwen3_5 model, _custom_ops
  Excluded: tests, benchmarks, docs, examples (not needed for reference)

Critical call chains now fully traceable:
  MoE: moe_topk_softmax_kernels.cuh → ixformer.h → fused_moe.cpp → layer
  GDN: qwen3_gated_delta_net_base.cpp → qwen3_5_gated_delta_net.cpp
  Attention: ixformer.h → xllm_paged_attention → attention.cpp

upstream_ref/xllm/examples/generate_beam_search.py

@@ -0,0 +1,29 @@
+# python examples/beam_search.py --model='/path/models/Qwen2-7B-Instruct' --devices='npu:0'
+# python beam_search.py --model='/path/models/Qwen2-7B-Instruct' --devices='npu:0,npu:1'
+
+from xllm import ArgumentParser, BeamSearchParams, LLM
+
+# Create an LLM.
+parser = ArgumentParser()
+llm = LLM(**vars(parser.parse_args()))
+
+beam_search_params = BeamSearchParams(
+    beam_width=2,
+    max_tokens=20,
+)
+
+outputs = llm.beam_search(
+    [
+     {"prompt": "Hello, my name is "},
+     {"prompt": "The president of the United States is "},
+     {"prompt": "The capital of France is "},
+     {"prompt": "The future of AI is "}
+    ],
+    params=beam_search_params,
+)
+
+for output in outputs:
+    generated_text = output.sequences[0].text
+    print(f"Generated text: {generated_text!r}")
+
+llm.finish()