ci: kill lingering next-server before restarting service to avoid EADDRINUSE

.gitea/workflows/deploy.yaml

@@ -48,7 +48,10 @@ jobs:
 
           # 8. Hard restart of the Systemd service to clear memory cache
           echo "Hard resetting systemd service..."
-          sudo systemctl stop atom-nexst.service
+          sudo systemctl stop atom-nexst.service || true
+          # Kill any lingering next-server processes holding port 3000 so the
+          # new instance can bind cleanly (systemctl stop may leave orphans).
+          pkill -f 'next-server' || true
           sudo systemctl start atom-nexst.service
 
           # Extra health check: Ensure the service is actually running