fix(sentinel): reduce resource churn from health flaps

Ignore health status changes in Sentinel push deduplication when the container lifecycle state is unchanged.

Scope stale resource checks to Sentinel servers whose heartbeat is stale, and avoid refreshing resource last_online_at on unchanged statuses.

tests/Feature/SentinelPushDeduplicationTest.php

@@ -138,6 +138,16 @@ it('dispatches the job when container state changes', function () use ($running)
     Queue::assertPushed(PushServerUpdateJob::class, 2);
 });
 
+it('ignores health status changes while container lifecycle state is unchanged', function () {
+    $healthy = [['name' => 'app-1', 'state' => 'running', 'health_status' => 'healthy']];
+    $unhealthy = [['name' => 'app-1', 'state' => 'running', 'health_status' => 'unhealthy']];
+
+    pushSentinel($this->token, sentinelPayload($healthy))->assertOk();
+    pushSentinel($this->token, sentinelPayload($unhealthy))->assertOk();
+
+    Queue::assertPushed(PushServerUpdateJob::class, 1);
+});
+
 it('ignores disk percentage changes (excluded from the hash)', function () use ($running) {
     pushSentinel($this->token, sentinelPayload($running(), diskPercentage: 42.0))->assertOk();
     pushSentinel($this->token, sentinelPayload($running(), diskPercentage: 88.0))->assertOk();