fix(network): skip worker IPv6 allocation when no workers exist

Previously, when creating a cluster with only control plane nodes
(worker_count = 0), Terraform would still allocate 1 unused worker
IPv6 primary IP. This caused "Primary IP limit exceeded" errors and
wasted resources.

Changes:
- Worker IPv6: Use total_worker_count directly (no fallback to 1)
- Worker private IPs: Remove unnecessary fallback for consistency
- Aligns worker IPv6 behavior with worker IPv4 allocation

Fixes #284

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <[email protected]>

Author: mrclrchtr

network.tf

@@ -97,7 +97,7 @@ resource "hcloud_primary_ip" "worker_ipv4" {
 }
 
 resource "hcloud_primary_ip" "worker_ipv6" {
-  count         = var.enable_ipv6 ? local.total_worker_count > 0 ? local.total_worker_count : 1 : 0
+  count         = var.enable_ipv6 ? local.total_worker_count : 0
   name          = "${local.cluster_prefix}worker-${count.index + 1}-ipv6"
   datacenter    = data.hcloud_datacenter.this.name
   type          = "ipv6"
@@ -139,6 +139,6 @@ locals {
     for index in range(var.control_plane_count > 0 ? var.control_plane_count : 1) : cidrhost(hcloud_network_subnet.nodes.ip_range, index + 101)
   ]
   worker_private_ipv4_list = [
-    for index in range(local.total_worker_count > 0 ? local.total_worker_count : 1) : cidrhost(hcloud_network_subnet.nodes.ip_range, index + 201)
+    for index in range(local.total_worker_count) : cidrhost(hcloud_network_subnet.nodes.ip_range, index + 201)
   ]
 }