feat(cookie-gate): 调度器恢复重跑接线与控制台健康自检

调度器维护循环挂载 CookieGateRecoveryManager，登录态恢复后按原业务日期自动
重跑被阻断工作流并飞书确认；工作流级 required_accounts 声明接入目录解析；
采集器抛 CookieExpiredError 退出 75；控制台新增 /api/health 自检端点与进程内
看门狗，配合 deploy/watchdog 双层保活；安装脚本同步。

tests/test_console.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import http.client
 import json
+import os
 import re
 import shutil
 import subprocess
@@ -311,7 +312,7 @@ def test_notification_config_persists_app_scoped_multi_recipient_routes(
     )
     initial = service.notification_config()
 
-    assert len(initial["people"]) == 16
+    assert len(initial["people"]) == 18
     assert initial["available_app_profiles"] == ["hermes-analyzer"]
     assert [app["profile"] for app in initial["apps"]] == ["hermes-analyzer"]
     wang = next(person for person in initial["people"] if person["id"] == "wang_yunlong")
@@ -321,7 +322,12 @@ def test_notification_config_persists_app_scoped_multi_recipient_routes(
     assert wang["bindings"]["hermes-analyzer"]["open_id"] == (
         "ou_7ad5fc8012e2f741afc5346e05ffd447"
     )
-    assert all(route["configured"] is False for route in initial["routes"])
+    stock_alert_route = next(
+        route
+        for route in initial["routes"]
+        if route["workflow_id"] == "product.online_stock_alert.daily"
+    )
+    assert stock_alert_route["configured"] is True
 
     routes = [dict(route) for route in initial["routes"]]
     daily = next(
@@ -1566,7 +1572,7 @@ def test_overview_exposes_only_verified_scheduled_runs_as_cancellable(
             "schema_version": 1,
             "updated_at": "2026-08-14T02:00:00+00:00",
             "jobs": {
-                "content.relogin.weekly": {
+                "content.notes_master.daily": {
                     "status": "running",
                     "pid": 4310,
                     "business_date": "2026-08-14",
@@ -1589,7 +1595,7 @@ def test_overview_exposes_only_verified_scheduled_runs_as_cancellable(
         workflow_id: str,
         _job: dict[str, Any],
     ) -> tuple[bool, Any, float | None]:
-        if workflow_id == "content.relogin.weekly":
+        if workflow_id == "content.notes_master.daily":
             return True, VerifiedProcess(), 1_800_000_000.0
         return False, None, None
 
@@ -1600,7 +1606,7 @@ def test_overview_exposes_only_verified_scheduled_runs_as_cancellable(
     ).overview()
     workflows = {item["id"]: item for item in overview["workflows"]}
 
-    verified = workflows["content.relogin.weekly"]["scheduler_job"]
+    verified = workflows["content.notes_master.daily"]["scheduler_job"]
     assert verified["status"] == "running"
     assert verified["cancellable"] is True
     assert "pid" not in verified
@@ -1614,7 +1620,7 @@ def test_service_cancels_verified_scheduled_run_and_finalizes_journal(
     monkeypatch: pytest.MonkeyPatch,
     console_settings: Settings,
 ) -> None:
-    workflow_id = "content.relogin.weekly"
+    workflow_id = "content.notes_master.daily"
     _write_scheduler_state(
         console_settings,
         {
@@ -1906,24 +1912,13 @@ def test_content_production_preflight_rejects_non_production_modes(
     relogin = next(
         item
         for item in catalog.workflows
-        if item.workflow_id == "content.relogin.weekly"
+        if item.workflow_id == "content.summary.weekly"
     )
 
     with pytest.raises(ConsoleRequestError, match=message):
         console_module._validate_content_production_runtime(relogin, {flag: "1"})
 
 
-def test_content_relogin_preflight_does_not_require_postgres() -> None:
-    catalog = WorkflowCatalog.load(PROJECT_ROOT / "config")
-    relogin = next(
-        item
-        for item in catalog.workflows
-        if item.workflow_id == "content.relogin.weekly"
-    )
-
-    console_module._validate_content_production_runtime(relogin, {})
-
-
 @pytest.mark.parametrize(
     "workflow_id",
     ["content.notes_master.daily", "content.metrics.daily"],
@@ -3297,7 +3292,7 @@ def test_http_notification_config_requires_revision_and_supports_multi_recipient
         assert status == HTTPStatus.OK
         _assert_security_headers(response_headers)
         config = json.loads(body)
-        assert len(config["people"]) == 16
+        assert len(config["people"]) == 18
         assert {
             route["workflow_id"] for route in config["routes"]
         } == set(console_module.NOTIFICATION_CAPABILITIES)
@@ -3616,3 +3611,82 @@ def test_douyin_negative_reviews_range_uses_single_run_with_env_dates(
         {"business_date": "2026-09-19"},
     )
     assert "business_date_end" not in single
+
+
+def test_api_health_reports_process_and_scheduler_state(
+    console_settings: Settings,
+) -> None:
+    with _running_server(console_settings, FakeConsoleLauncher()) as server:
+        status, _headers, content = _http_request(server, "GET", "/api/health")
+        assert status == HTTPStatus.OK
+        payload = json.loads(content)
+        assert payload["ok"] is True
+        assert payload["pid"] == os.getpid()
+        assert payload["uptime_seconds"] >= 0
+        # 测试环境没有调度器实例锁，health 只报状态不做任何拉起。
+        assert payload["scheduler_running"] is False
+        assert payload["scheduler_pid"] is None
+
+
+def test_api_health_requires_no_token(console_settings: Settings) -> None:
+    # 自检探针端点供进程内看门狗与保活脚本使用，必须绕过访问令牌。
+    with _running_server(
+        console_settings, FakeConsoleLauncher(), token="x" * 24
+    ) as server:
+        status, _headers, _content = _http_request(server, "GET", "/api/health")
+        assert status == HTTPStatus.OK
+
+
+def test_probe_console_health_against_real_server(
+    console_settings: Settings,
+) -> None:
+    with _running_server(console_settings, FakeConsoleLauncher()) as server:
+        assert console_module._probe_console_health(server.server_port, 2.0) is True
+        assert console_module._probe_console_health(1, 0.2) is False
+
+
+def test_console_watchdog_exits_after_consecutive_probe_failures() -> None:
+    stopped = threading.Event()
+    exits: list[int] = []
+    probes: list[int] = []
+
+    def always_failing(port: int, _timeout: float) -> bool:
+        probes.append(port)
+        return False
+
+    console_module._console_watchdog(
+        8765,
+        interval_seconds=0.01,
+        failure_limit=3,
+        probe_timeout_seconds=0.01,
+        stopped=stopped,
+        probe=always_failing,
+        exit_process=exits.append,
+        sleep=lambda _seconds: None,
+    )
+
+    assert exits == [console_module.CONSOLE_WATCHDOG_EXIT_CODE]
+    assert probes == [8765, 8765, 8765]
+
+
+def test_console_watchdog_resets_failure_count_after_successful_probe() -> None:
+    stopped = threading.Event()
+    exits: list[int] = []
+    results = iter([False, False, True, False, False, False])
+
+    def flaky_probe(_port: int, _timeout: float) -> bool:
+        return next(results)
+
+    console_module._console_watchdog(
+        8765,
+        interval_seconds=0.01,
+        failure_limit=3,
+        probe_timeout_seconds=0.01,
+        stopped=stopped,
+        probe=flaky_probe,
+        exit_process=exits.append,
+        sleep=lambda _seconds: None,
+    )
+
+    # 成功一次清零计数：2 次失败 + 1 次成功 + 3 次失败后才退出。
+    assert exits == [console_module.CONSOLE_WATCHDOG_EXIT_CODE]