fix: clear in-flight after post_result to prevent double-pings, revert poll to 50ms

apps/monitor/src/main.rs

@@ -41,6 +41,6 @@ async fn main() -> Result<()> {
             Ok(n) => { if n > 0 { info!("Spawned {n} checks"); } },
             Err(e) => error!("Check cycle failed: {e}"),
         }
-        sleep(Duration::from_millis(500)).await;
+        sleep(Duration::from_millis(50)).await;
     }
 }

apps/monitor/src/runner.rs

@@ -61,11 +61,12 @@ pub async fn fetch_and_run(
                     meta: None,
                 },
             };
-            // Remove from in-flight before posting so a fast next cycle can pick it up
-            in_flight.lock().await.remove(&monitor.id);
+            // Post result first, then clear in-flight — this prevents the next
+            // poll from picking up the monitor again before the ping is persisted.
             if let Err(e) = post_result(&client, &coordinator_url, &token, result).await {
                 warn!("Failed to post result for {}: {e}", monitor.id);
             }
+            in_flight.lock().await.remove(&monitor.id);
         });
     }