feat: initial vllm-npu-plugin for Ascend NPU adaptation

- NPUPlatform: device management, HCCL process group, config adaptation
- AscendAttentionBackend: npu_fusion_attention (prefill) + npu_incre_flash_attention (decode)
- NPUCommunicator: HCCL-based distributed communication
- NPUWorker: NPU device init, memory profiling
- Custom ops: SiluAndMul, RMS norm, rotary embedding
- Plugin registered via vllm.platform_plugins entry point

Based on vllm-ascend official pattern, targeting Ascend 910B

setup.py

@@ -0,0 +1,24 @@
+"""
+vllm-npu-plugin: Ascend NPU platform plugin for vLLM v0.11.0.
+
+Registers as an out-of-tree platform plugin via the
+``vllm.platform_plugins`` entry-point group.
+"""
+
+from setuptools import find_packages, setup
+
+setup(
+    name="vllm-npu-plugin",
+    version="0.1.0",
+    description="Ascend NPU platform plugin for vLLM v0.11.0",
+    packages=find_packages(),
+    python_requires=">=3.9",
+    install_requires=[
+        # vllm must already be installed (v0.11.0)
+    ],
+    entry_points={
+        "vllm.platform_plugins": [
+            "npu = vllm_npu:register",
+        ],
+    },
+)