fix: ignore default node_pool in cluster lifecycle to prevent recreation

terraform/doks.tf

@@ -5,13 +5,17 @@ resource "digitalocean_kubernetes_cluster" "main" {
   region  = var.region
   version = var.k8s_version
 
-  # Default pool — scaled to 0, workloads run on system/customer pools
+  # Required by provider but managed externally — do not let Terraform recreate
   node_pool {
     name       = "workers"
     size       = "s-2vcpu-4gb"
     node_count = 0
   }
 
+  lifecycle {
+    ignore_changes = [node_pool]
+  }
+
   tags = ["lunarfront", "k8s"]
 }