refactor: resource-based scaling only, remove health-check triggered scaling

Scale up: only if resource constrained (CPU/memory)
Scale down: only if NOT resource constrained AND game load low
Remove: triggering scale-up on unexpected instance failures
Keep: health monitoring (mark dead, delete pod, failover games) but no scaling

Prevents cascade scaling from transient health check failures.

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

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

@@ -182,16 +182,14 @@ class AutoScaler:
           val hasHighCpuOrMemory  = constrainedInstance.isDefined
 
           log.infof(
-            "Scale check: instances=%d avgLoad=%.1f scaleUpAt=%.1f scaleDownAt=%.1f resourceConstrained=%s",
+            "Scale check: instances=%d avgLoad=%.1f resourceConstrained=%s",
             instances.size,
             avgLoad,
-            scaleUpLoad,
-            scaleDownLoad,
             constrainedInstance.map(_.instanceId).getOrElse("none"),
           )
 
-          if avgLoad > scaleUpLoad || hasHighCpuOrMemory then scaleUp()
-          else if avgLoad < scaleDownLoad && instances.size > config.scaleMinReplicas
+          if hasHighCpuOrMemory then scaleUp()
+          if !hasHighCpuOrMemory && avgLoad < scaleDownLoad && instances.size > config.scaleMinReplicas
           then scaleDown()
 
   private def patchRolloutReplicas(

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

@@ -88,9 +88,7 @@ class HealthMonitor:
     if evicted.nonEmpty then
       log.warnf("Evicted %d stale instances: %s", evicted.size, evicted.mkString(", "))
       evicted.foreach(deleteK8sPod)
-      val unexpectedEvictions = evicted.filterNot(autoScaler.isDrainingForScaleDown)
       evicted.foreach(autoScaler.clearDraining)
-      if unexpectedEvictions.nonEmpty then autoScaler.scaleUp()
     val instances = instanceRegistry.getAllInstances
     val failed = instances.collect { inst =>
       val isHealthy = checkHealth(inst.instanceId)
@@ -101,8 +99,6 @@ class HealthMonitor:
         Some(inst.instanceId)
       else None
     }.flatten
-    val unexpectedFailures = failed.filterNot(autoScaler.isDrainingForScaleDown)
-    if unexpectedFailures.nonEmpty then autoScaler.scaleUp()
 
   private def checkHealth(instanceId: String): Boolean =
     val redisHealthy = checkRedisHeartbeat(instanceId)