fix: avoid server replacement; install tailscale via Ansible

.gitea/workflows/deploy.yml

@@ -16,7 +16,6 @@ env:
   TF_VAR_s3_secret_key: ${{ secrets.S3_SECRET_KEY }}
   TF_VAR_s3_endpoint: ${{ secrets.S3_ENDPOINT }}
   TF_VAR_s3_bucket: ${{ secrets.S3_BUCKET }}
-  TF_VAR_tailscale_auth_key: ${{ secrets.TAILSCALE_AUTH_KEY }}
   TF_VAR_tailscale_tailnet: ${{ secrets.TAILSCALE_TAILNET }}
 
 jobs:
@@ -175,6 +174,7 @@ jobs:
         run: |
           ansible-playbook site.yml \
             -e "hcloud_token=${{ secrets.HCLOUD_TOKEN }}" \
+            -e "tailscale_auth_key=${{ secrets.TAILSCALE_AUTH_KEY }}" \
             -e "cluster_name=k8s-cluster"
         env:
           ANSIBLE_HOST_KEY_CHECKING: "False"