fix: revert Dockerfile hacks, restart CoreDNS to fix DNS resolution

Root cause: CoreDNS pods were failing to forward external DNS queries
(registry.npmjs.org.celestium.life -> no next plugin). Restarted
CoreDNS deployment. Reverted Dockerfile to original clean state.

.woodpecker/build-3.yml

@@ -197,11 +197,9 @@ steps:
       registry: registry.celestium.life
       buildx_image: registry.celestium.life/dockerhub-cache/moby/buildkit:buildx-stable-1
       add_host: registry.celestium.life:10.1.1.12
-      buildx_flags: --driver-opt network=host --allow network.host
+      buildx_flags: --driver-opt network=host
       insecure: true
       buildkit_config: |
-        [worker.oci]
-          networkMode = "host"
         [registry."registry.celestium.life"]
           insecure = true
         [registry."docker.io"]

frontend/Dockerfile

@@ -1,4 +1,3 @@
-# syntax=docker/dockerfile:1
 # Stage 1: Build
 # Base images: uses Harbor proxy cache in CI, falls back to Docker Hub externally
 FROM registry.celestium.life/dockerhub-cache/library/node:24-alpine AS build