refactor: 重构 cache 和 inference 参数体系，分离存储与分配

- 合并 GenerationRequest/GenerationParams，统一 max_tokens 参数名
- PagePool/PrefixCache 分离为 Allocator + PrefixCache + PagePool
- 拆分 KV 存储为独立 Storage 类，PagedCache → KVCache，CacheView → KvcacheView
- Allocator.inc_ref 移除 LRU 防止竞争，Storage.write 增加负页防御
- Allocator/PrefixCache/TaskTable 加 threading.Lock 保证线程安全
- server.py uvicorn.run 改为传 app 对象修复导入错误
- benchmark.py 适配 KVCache 新 API

astrai/inference/api/protocol.py

@@ -14,7 +14,7 @@ from typing import Any, Dict, List, Optional, Union
 from fastapi.responses import StreamingResponse
 from pydantic import BaseModel
 
-from astrai.inference.engine import GenerationParams, InferenceEngine
+from astrai.inference.engine import InferenceEngine
 
 
 def _sse_event(data: Dict[str, Any], event: Optional[str] = None) -> str:
@@ -143,13 +143,13 @@ class ProtocolHandler(ABC):
             prompt_tokens=self._count_prompt_tokens(),
         )
 
-        params = GenerationParams(
+        agen = self.engine.generate_async(
+            prompt=self.build_prompt(),
             max_tokens=self.request.max_tokens,
             temperature=self.request.temperature,
             top_p=self.request.top_p,
             top_k=self.request.top_k,
         )
-        agen = self.engine.generate_async(prompt=self.build_prompt(), params=params)
 
         if self.request.stream:
             return self._handle_stream(agen, ctx)

astrai/inference/api/server.py

@@ -160,8 +160,7 @@ def run_server(
         "max_batch_size": max_batch_size,
     }
     uvicorn.run(
-        "astrai.inference.server:app",
+        app,
         host=host,
         port=port,
-        reload=reload,
     )