refactor: 增加内存保护措施，精简启动任务

- 流式响应不再存储完整文本，仅记录长度用于 token 估算
- complete_response 文本累积增加 64KB 上限保护
- 通知缓冲区(email/webhook)增加溢出保护，超限丢弃旧通知
- 熔断器淘汰策略改进，全部 open/half-open 时按最旧失败时间淘汰
- 移除启动时清理任务和统计聚合回填，减少启动负担
- token 估算简化为基于长度的方法，避免持有完整文本

src/plugins/notification/email.py

@@ -55,6 +55,7 @@ class EmailNotificationPlugin(NotificationPlugin):
 
         # 缓冲配置
         self._buffer: list[Notification] = []
+        self._buffer_max_size = config.get("buffer_max_size", 500) if config else 500
         self._lock = asyncio.Lock()
         self._flush_task: asyncio.Task[None] | None = None
 
@@ -285,6 +286,12 @@ class EmailNotificationPlugin(NotificationPlugin):
         """
         # 添加到缓冲区
         async with self._lock:
+            # 缓冲区溢出保护：丢弃最旧的通知
+            if len(self._buffer) >= self._buffer_max_size:
+                drop_count = len(self._buffer) - self._buffer_max_size + 1
+                del self._buffer[:drop_count]
+                logger.warning("Email 通知缓冲区溢出，丢弃 {} 条旧通知", drop_count)
+
             self._buffer.append(notification)
 
             # 如果是严重通知，立即发送

src/plugins/notification/webhook.py

@@ -52,6 +52,7 @@ class WebhookNotificationPlugin(NotificationPlugin):
 
         # 缓冲配置
         self._buffer: list[Notification] = []
+        self._buffer_max_size = config.get("buffer_max_size", 500) if config else 500
         self._lock = asyncio.Lock()
         self._session: aiohttp.ClientSession | None = None
         self._flush_task = None
@@ -199,6 +200,12 @@ class WebhookNotificationPlugin(NotificationPlugin):
         """
         # 添加到缓冲区
         async with self._lock:
+            # 缓冲区溢出保护：丢弃最旧的通知
+            if len(self._buffer) >= self._buffer_max_size:
+                drop_count = len(self._buffer) - self._buffer_max_size + 1
+                del self._buffer[:drop_count]
+                logger.warning("Webhook 通知缓冲区溢出，丢弃 {} 条旧通知", drop_count)
+
             self._buffer.append(notification)
 
             # 如果是严重通知，立即发送