fix: don't block event loop during scale-down drain

Scale-down was calling failoverService.onInstanceStreamDropped synchronously
and waiting for it to complete. Failover retries for up to 30s waiting for
healthy instances, which blocks the Quarkus event loop thread.

This caused:
- Event loop blocked for 15+ seconds
- Redis health checks timing out (also on event loop)
- Scale-down operations failing

Fix: Trigger drain asynchronously without waiting. Scale-down proceeds
immediately while drain happens in background.

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>

modules/coordinator/src/main/scala/de/nowchess/coordinator/service/AutoScaler.scala

@@ -134,12 +134,16 @@ class AutoScaler:
       .minByOption(_.subscriptionCount)
 
     underloadedInstance.foreach { inst =>
-      log.infof("Draining instance %s before scale-down", inst.instanceId)
+      log.infof("Marking instance %s for drain before scale-down", inst.instanceId)
       failoverService
         .onInstanceStreamDropped(inst.instanceId)
         .subscribe()
         .`with`(
-          _ =>
+          _ => log.debugf("Instance %s drained for scale-down", inst.instanceId),
+          ex => log.warnf(ex, "Drain failed for %s, proceeding with scale-down", inst.instanceId),
+        )
+    }
+
     kubeClientOpt match
       case None =>
         log.warn("Kubernetes client not available, cannot scale")
@@ -179,11 +183,4 @@ class AutoScaler:
         catch
           case ex: Exception =>
             meterRegistry.counter("nowchess.coordinator.scale.failures", "direction", "down").increment()
-                    log.warnf(ex, "Failed to scale down %s", config.k8sRolloutName),
+            log.warnf(ex, "Failed to scale down %s", config.k8sRolloutName)
-          ex =>
-            meterRegistry.counter("nowchess.coordinator.scale.failures", "direction", "down").increment()
-            log.warnf(ex, "Failed to drain instance %s before scale-down", inst.instanceId),
-        )
-    }
-
-    if underloadedInstance.isEmpty then log.warn("No healthy instances found for scale-down")