fix: make backend DNS check optional when getent is unavailable

Add getent availability check before attempting DNS resolution in wait_for_backend_dns function. Fall back to allowing curl connectivity check in main loop when getent is not available, supporting Alpine-based images that don't ship getent by default.

deploy/scripts/gateway-entrypoint.sh

@@ -57,12 +57,18 @@ NFT_CONF="/var/lib/nexavpn/nftables.generated.conf"
 mkdir -p /etc/wireguard
 
 wait_for_backend_dns() {
-  if getent hosts "${BACKEND_HOST}" >/dev/null 2>&1; then
-    return 0
+  if command -v getent >/dev/null 2>&1; then
+    if getent hosts "${BACKEND_HOST}" >/dev/null 2>&1; then
+      return 0
+    fi
+
+    echo "Backend host ${BACKEND_HOST} is not resolvable yet."
+    return 1
   fi
 
-  echo "Backend host ${BACKEND_HOST} is not resolvable yet."
-  return 1
+  # Alpine-based images may not ship getent. In that case let curl be the
+  # source of truth for connectivity and retry in the main loop on failure.
+  return 0
 }
 
 apply_bundle() {