feat(alerting): add Telegram callback ingress sensor

apps/api/src/main.py

@@ -339,6 +339,11 @@ async def lifespan(_app: FastAPI) -> AsyncGenerator[None, None]:
     # OpenClaw (192.168.0.188) 是唯一的 Polling 實例
     # AWOOOI API 只負責發送訊息，不接收
     telegram_gw = get_telegram_gateway()
+    callback_receipt_loaded = await telegram_gw.refresh_callback_ingress_receipt()
+    logger.info(
+        "telegram_callback_ingress_receipt_bootstrap",
+        durable_receipt_loaded=callback_receipt_loaded,
+    )
     if settings.TELEGRAM_ENABLE_POLLING:
         await telegram_gw.start_long_polling()
         logger.info("telegram_long_polling_started")
@@ -1400,6 +1405,16 @@ app.include_router(
 @app.get("/metrics", include_in_schema=False)
 async def prometheus_metrics() -> Response:
     """Prometheus metrics endpoint for alerting"""
+    # Callback ingress receipts are shared in Redis while API metrics are
+    # process-local. Refresh before every scrape so a NodePort that alternates
+    # across replicas cannot expose stale/false-green transport evidence.
+    try:
+        await get_telegram_gateway().refresh_callback_ingress_receipt()
+    except Exception as exc:
+        logger.warning(
+            "prometheus_metrics_telegram_callback_receipt_error",
+            error_type=type(exc).__name__,
+        )
     # 2026-05-19 Codex — T85 Alert Chain DB evidence refresh.
     # record_alert_chain_success() 是 process-local gauge；部署後第一個 scrape
     # 可能尚未收到新 webhook，導致 smoke test 誤判 metric 不存在。