fix: refresh Redis TTL on instance heartbeat to prevent false DEAD marking

Instances were being incorrectly marked DEAD because their Redis key TTL was
not being refreshed on heartbeat. HealthMonitor.checkRedisHeartbeat() checks
pttl > 0, which fails when the TTL expires even if the instance is alive and
sending regular heartbeats.

Now pexpire(key, heartbeatTtl) is called on each heartbeat to keep the key
alive. Prevents scaling messages from undercounting healthy instances.

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

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

@@ -9,6 +9,7 @@ import scala.jdk.CollectionConverters.*
 import scala.compiletime.uninitialized
 import com.fasterxml.jackson.databind.ObjectMapper
 import de.nowchess.coordinator.dto.InstanceMetadata
+import de.nowchess.coordinator.config.CoordinatorConfig
 import java.util.concurrent.ConcurrentHashMap
 import java.time.{Duration, Instant}
 import io.micrometer.core.instrument.{Gauge, MeterRegistry}
@@ -27,6 +28,9 @@ class InstanceRegistry:
 
   @Inject
   private var meterRegistry: MeterRegistry = uninitialized
+
+  @Inject
+  private var config: CoordinatorConfig = uninitialized
   // scalafix:on DisableSyntax.var
 
   private val log       = Logger.getLogger(classOf[InstanceRegistry])
@@ -95,7 +99,13 @@ class InstanceRegistry:
                 metadata.subscriptionCount,
                 metadata.state,
               )
-            Uni.createFrom().item(())
+            val ttlMs = config.heartbeatTtl.toMillis
+            redis
+              .key(classOf[String])
+              .pexpire(key, ttlMs)
+              .map(_ => ())
+              .onFailure()
+              .recoverWithItem(())
           }
         catch
           case ex: Exception =>