Fix Destroy being immediately undone by dashboard auto-redeploy

Dashboard was auto-deploying any missing stack on every page load.
This stomped on the Destroy action. Stack creation only happens at
activation and via explicit Start — not on dashboard render.

Made-with: Cursor

docker/ss-atlas/internal/handlers/dashboard.go

@@ -38,14 +38,6 @@ func (a *App) handleDashboard(w http.ResponseWriter, r *http.Request) {
 		if err != nil {
 			log.Printf("dashboard: stack check failed for %s: %v", remoteUser, err)
 		}
-		if !exists {
-			log.Printf("dashboard: deploying missing stack %s", stackName)
-			if err := a.swarm.DeployStack(stackName, remoteUser, a.cfg.TraefikDomain); err != nil {
-				log.Printf("dashboard: stack deploy failed for %s: %v", remoteUser, err)
-			} else {
-				exists = true
-			}
-		}
 		stackDeployed = exists
 		if exists {
 			replicas, _ := a.swarm.GetWebReplicas(stackName)

docker/ss-atlas/templates/stack-template.yml

@@ -19,9 +19,6 @@
 # Each customer gets their stack at: https://{{.Subdomain}}.{{.Domain}}
 # Access is restricted to the owning user via Authelia forward-auth.
 # =============================================================================
-
-version: "3.8"
-
 services:
   web:
     image: louislam/uptime-kuma:1