feat: 流式空闲超时、健康监控查询优化、限流桶内存上限与维护清理修复

Close #233

Co-authored-by: AAEE86 <ppk0227@hotmail.com>

- cli_monitor_mixin: 引入 STREAM_IDLE_TIMEOUT_SECONDS（可通过环境变量配置），
  流传输开始后若超出空闲窗口无新 chunk 则提前取消并返回 504，避免长时间挂起
- stream_context: 新增 managed_recorded_bodies 上下文管理器，确保 chunks 在
  telemetry 完成后及时释放；stream_telemetry 使用该接口统一管理 response body 构建
- health endpoint: 将状态聚合改为 GROUP BY 直接统计，事件列表按 api_format
  单独查询，避免单次 limit 拉取大量记录导致的遗漏与性能问题；同时过滤不活跃
  provider/endpoint，与公开健康接口保持一致
- endpoint health service: 修正时间线数据按 endpoint_id 而非 key_id 聚合
- token_bucket: 引入 max_buckets/bucket_expiry 上限与定时清理，防止内存无限增长；
  修复 refill_rate=0 时 get_reset_time 除零异常；新增 _is_unlimited_rate_limit 判断
- maintenance_scheduler: 调整清理顺序（先删整行再按窗口清理），新增 newer_than
  边界参数，避免同一行在同一轮中被重复改写
- sync_execute: 新增 create_pending_usage 开关，允许已预创建记录的调用方跳过重复创建
- quota_reader / provider_ops balance: 小幅修复与健壮性提升
- Dockerfile: 添加 MALLOC_ARENA_MAX=2 环境变量以降低 gunicorn worker RSS
- 补充相关测试覆盖

tests/api/handlers/base/test_stream_telemetry.py

@@ -1,11 +1,15 @@
 from __future__ import annotations
 
+import os
 import time
 from types import SimpleNamespace
+from typing import Any
 from unittest.mock import AsyncMock
 
 import pytest
 
+os.environ.setdefault("JWT_SECRET_KEY", "test-secret-key")
+
 from src.api.handlers.base import stream_telemetry as stream_telemetry_module
 from src.api.handlers.base.stream_context import StreamContext
 from src.api.handlers.base.stream_telemetry import StreamTelemetryRecorder
@@ -64,3 +68,69 @@ async def test_record_stream_stats_estimates_tokens_for_failed_partial_stream(
     assert ctx.input_tokens > 0
     assert ctx.output_tokens == max(1, len("partial output") // 4)
     recorder._dispatch_record.assert_awaited_once()  # type: ignore[attr-defined]
+
+
+@pytest.mark.asyncio
+async def test_record_stream_stats_releases_parsed_chunks_after_dispatch(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    recorder = StreamTelemetryRecorder(
+        request_id="req-release",
+        user_id="1",
+        api_key_id="2",
+        client_ip="127.0.0.1",
+        format_id="openai:chat",
+    )
+    recorder._get_telemetry_writer = AsyncMock(  # type: ignore[method-assign]
+        return_value=SimpleNamespace(include_bodies=True)
+    )
+    recorder._update_candidate_status = AsyncMock()  # type: ignore[method-assign]
+
+    dispatch_payloads: list[dict[str, Any]] = []
+
+    async def _capture_dispatch(*args: Any, **kwargs: Any) -> None:
+        dispatch_payloads.append(
+            {
+                "response_body": args[5],
+                "client_response_body": kwargs.get("client_response_body"),
+            }
+        )
+
+    recorder._dispatch_record = _capture_dispatch  # type: ignore[method-assign]
+
+    ctx = StreamContext(
+        model="test-model",
+        api_format="openai:chat",
+        request_id="req-release",
+        user_id=1,
+        api_key_id=2,
+    )
+    ctx.provider_name = "test-provider"
+    ctx.parsed_chunks.extend([{"type": "chunk-1"}, {"type": "chunk-2"}])
+    ctx.data_count = 2
+    ctx.chunk_count = 2
+
+    monkeypatch.setattr(stream_telemetry_module, "get_db", lambda: iter([_DummyDb()]))
+    monkeypatch.setattr(
+        stream_telemetry_module.SystemConfigService,
+        "should_log_body",
+        lambda _db: True,
+    )
+    monkeypatch.setattr(stream_telemetry_module.config, "stream_stats_delay", 0)
+
+    await recorder.record_stream_stats(
+        ctx,
+        original_headers={},
+        original_request_body={"input": [{"content": "hello world"}]},
+        start_time=time.time(),
+    )
+
+    assert len(dispatch_payloads) == 1
+    response_body = dispatch_payloads[0]["response_body"]
+    assert response_body["chunks"] == [{"type": "chunk-1"}, {"type": "chunk-2"}]
+    assert response_body["metadata"]["stream"] is True
+    assert response_body["metadata"]["total_chunks"] == 2
+    assert response_body["metadata"]["data_count"] == 2
+    assert dispatch_payloads[0]["client_response_body"] is None
+    assert ctx.parsed_chunks == []
+    assert ctx.provider_parsed_chunks == []