fix: restore changedetection and flaresolverr services

- changedetection: increase memory from 64Mi to 256Mi/512Mi (was OOMKilling),
  set replicas back to 1
- flaresolverr: re-enable with replicas=1, increase memory limit to 1Gi
  (needed by book-search for Cloudflare bypass)

stacks/changedetection/main.tf

@@ -99,7 +99,7 @@ resource "kubernetes_deployment" "changedetection" {
   }
   spec {
     # Disabled: chronic OOM at 64Mi limit, not worth the memory cost to increase
-    replicas = 0
+    replicas = 1
     strategy {
       type = "Recreate"
     }
@@ -171,10 +171,10 @@ resource "kubernetes_deployment" "changedetection" {
           resources {
             requests = {
               cpu    = "15m"
-              memory = "64Mi"
+              memory = "256Mi"
             }
             limits = {
-              memory = "64Mi"
+              memory = "512Mi"
             }
           }
         }

stacks/servarr/flaresolverr/main.tf

@@ -14,7 +14,7 @@ resource "kubernetes_deployment" "flaresolverr" {
     }
   }
   spec {
-    replicas = 0 # Scaled down — headless Chrome with no effective resource limits causes node OOM
+    replicas = 1
     selector {
       match_labels = {
         app = "flaresolverr"
@@ -37,7 +37,7 @@ resource "kubernetes_deployment" "flaresolverr" {
               memory = "512Mi"
             }
             limits = {
-              memory = "512Mi"
+              memory = "1Gi"
             }
           }
           port {