fix(alerts): Steward's own outage no longer reads as every agent going silent (#5441)

After Steward or its database was down, every agent's age and every
series' last sample were old, so all agent-offline rules fired and all
rules went no-data at once, clearing a minute later. With per-account
alert email that would have mailed everyone a dozen notices per restart.

A periodic job now ticks a Watch (steward/core/watch.py) once it has
reached the database; the Watch says since when runs have been unbroken
(process start, or the end of a gap over 120 s). seconds_since_seen and
the no-data sweep count silence from no earlier than that, so a source
must stay quiet while Steward is watching before it counts as silent.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

steward/core/watch.py

@@ -0,0 +1,43 @@
+"""Since when Steward has been watching without a break (#5441).
+
+Silence is measured by Steward: an agent's age is now minus its last report,
+a series is silent once no sample has arrived for a while. While Steward (or
+its database) is down nothing arrives, so on its return every agent and every
+series looks silent at once, and every agent-offline and no-data alert used to
+fire together for an outage that was Steward's own.
+
+A periodic job ticks its Watch each run, once it has reached the database.
+The Watch answers "since when have I been running without a break": since
+the process started, or since the end of a gap longer than GAP_S between
+ticks (the database was down, the loop stalled). Silence is then counted from
+no earlier than that, so a source has to stay quiet while Steward is
+watching before it counts as silent.
+"""
+from __future__ import annotations
+
+from datetime import datetime
+
+# Two missed 60 s ticks. A gap this long means Steward was not watching; a
+# shorter one only delays a real alert by the length of the gap.
+GAP_S = 120.0
+
+
+class Watch:
+    def __init__(self) -> None:
+        self.since: datetime | None = None
+        self.last: datetime | None = None
+
+    def tick(self, now: datetime) -> datetime:
+        """Record a run at `now`; return since when runs have been unbroken."""
+        if self.since is None or self.last is None or (now - self.last).total_seconds() > GAP_S:
+            self.since = now
+        self.last = now
+        return self.since
+
+
+def watch(app, name: str) -> Watch:
+    """The app's Watch for one periodic job, made on first use."""
+    watches = getattr(app, "_watches", None)
+    if watches is None:
+        watches = app._watches = {}
+    return watches.setdefault(name, Watch())