refactor(alert): refresh stale_schedule docs; log tick schedule errors; add mode-change + never-backed-up tests

internal/alert/engine.go

@@ -196,11 +196,9 @@ func (e *Engine) handleHostOnline(ctx context.Context, hostID string) {
 // tick is the 60-second sweep. Responsibilities:
 //  1. Re-evaluate agent_offline for every offline host that may have
 //     crossed the floor between explicit events.
-//  2. Stale-schedule detection — declared in the spec but intentionally
-//     left as a no-op in v1. The precise "expected to have fired but
-//     didn't" trigger requires a store helper that lands in a later
-//     task. The KindStaleSchedule constant is exported so UI code can
-//     reference the tag string today.
+//  2. Stale-schedule detection for intermittent hosts — raises
+//     stale_schedule when LastBackupAt is older than 7 days and the
+//     host has an enabled schedule. Always-on hosts are excluded.
 func (e *Engine) tick(ctx context.Context, now time.Time) {
 	// User-management cleanup piggy-backed here for now. Setup tokens
 	// have a 1h expiry; the alert engine tick is the cheapest existing
@@ -232,7 +230,11 @@ func (e *Engine) tick(ctx context.Context, now time.Time) {
 				continue
 			}
 			hasEnabled, err := e.hostHasEnabledSchedule(ctx, h.ID)
-			if err != nil || !hasEnabled {
+			if err != nil {
+				slog.Warn("alert: tick list schedules", "host_id", h.ID, "err", err)
+				continue
+			}
+			if !hasEnabled {
 				continue
 			}
 			e.raiseAndNotify(ctx, h.ID, KindStaleSchedule, "", "warning",