refactor: 引入 safe_create_task 防止后台任务被 GC 回收，降低默认连接池和 worker 数量

- 新增 safe_create_task 统一替代裸 asyncio.create_task，通过全局集合持有 task 引用
- 默认 worker 数量从 4 降为 1，HTTP 连接池总预算从 800 降为 200
- 为 health_cache 和 affinity_manager 内存缓存增加上限淘汰机制
- MemoryCachePlugin 支持延迟启动清理任务
- gunicorn when_ready 增加 gc.collect() 并记录 post_worker_init RSS

tests/services/test_user_service_delete_user.py

@@ -141,7 +141,7 @@ def test_delete_user_precleans_large_tables_before_final_delete(
         "src.services.user.service.UserCacheService.invalidate_user_cache",
         invalidate_user_cache,
     )
-    monkeypatch.setattr("src.services.user.service.asyncio.create_task", create_task)
+    monkeypatch.setattr("src.services.user.service.safe_create_task", create_task)
 
     assert UserService.delete_user(db, "user-3") is True
 

tests/unit/test_runtime_http_pool_defaults.py

@@ -0,0 +1,34 @@
+import pytest
+
+from src.config.settings import Config
+
+
+def test_config_defaults_to_single_worker_and_capped_http_pool(
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    for key in (
+        "WEB_CONCURRENCY",
+        "GUNICORN_WORKERS",
+        "HTTP_MAX_CONNECTIONS",
+        "HTTP_KEEPALIVE_CONNECTIONS",
+    ):
+        monkeypatch.delenv(key, raising=False)
+
+    cfg = Config()
+
+    assert cfg.worker_processes == 1
+    assert cfg.http_max_connections == 200
+    assert cfg.http_keepalive_connections == 60
+
+
+def test_config_scales_http_pool_down_for_multi_worker(monkeypatch: pytest.MonkeyPatch) -> None:
+    monkeypatch.setenv("GUNICORN_WORKERS", "2")
+    monkeypatch.delenv("WEB_CONCURRENCY", raising=False)
+    monkeypatch.delenv("HTTP_MAX_CONNECTIONS", raising=False)
+    monkeypatch.delenv("HTTP_KEEPALIVE_CONNECTIONS", raising=False)
+
+    cfg = Config()
+
+    assert cfg.worker_processes == 2
+    assert cfg.http_max_connections == 100
+    assert cfg.http_keepalive_connections == 30