refactor: 统一类型注解风格并优化代码结构

- 将裸 except 改为显式 Exception 捕获
- 用列表推导式替换冗余 for 循环
- 为类属性添加 ClassVar 注解
- 统一 Union/Optional 写法为 |
- 移除未使用的导入
- 修复 SQLAlchemy 空值比较语法
- 优化字符串拼接与字典更新逻辑
- 补充缺失的 noqa 注释与异常链

BREAKING CHANGE: 所有插件基类的类级字段现要求显式 ClassVar 注解，自定义插件需同步更新

src/chat/knowledge/embedding_store.py

@@ -306,10 +306,8 @@ class EmbeddingStore:
 
     def save_to_file(self) -> None:
         """保存到文件"""
-        data = []
         logger.info(f"正在保存{self.namespace}嵌入库到文件{self.embedding_file_path}")
-        for item in self.store.values():
-            data.append(item.to_dict())
+        data = [item.to_dict() for item in self.store.values()]
         data_frame = pd.DataFrame(data)
 
         if not os.path.exists(self.dir):

src/chat/knowledge/utils/dyn_topk.py

@@ -15,15 +15,14 @@ def dyn_select_top_k(
     # 归一化
     max_score = sorted_score[0][1]
     min_score = sorted_score[-1][1]
-    normalized_score = []
-    for score_item in sorted_score:
-        normalized_score.append(
-            (
-                score_item[0],
-                score_item[1],
-                (score_item[1] - min_score) / (max_score - min_score),
-            )
+    normalized_score = [
+        (
+            score_item[0],
+            score_item[1],
+            (score_item[1] - min_score) / (max_score - min_score),
         )
+        for score_item in sorted_score
+    ]
 
     # 寻找跳变点：score变化最大的位置
     jump_idx = 0