fix(apps): explicit forte-drop namespace (sync-wave -1, Prune=false)

Codex review: the apps overlay applies namespaced resources
(keycloak-client Secrets, forte-drop-secrets, PDB) to forte-drop, but
no base created the namespace — first sync on a fresh cluster raced
ahead of the Applications' CreateNamespace and failed with
'namespaces forte-drop not found' until a retry.

Add an explicit Namespace at sync-wave -1 so it exists before the
wave-0 namespaced resources (covers both web + mcp bases via the
shared parent). Prune=false keeps removing a base from cascade-
deleting the namespace + postgres data + the other deployment.

apps/base/forte-drop/kustomization.yaml

@@ -1,6 +1,7 @@
 apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 resources:
+- namespace.yaml
 - forte-drop.yaml
 - keycloak-client-forte-drop.yaml
 - forte-drop-pdb.yaml

apps/base/forte-drop/namespace.yaml

@@ -0,0 +1,17 @@
+# Owns the forte-drop namespace shared by the web + mcp deployments and the
+# postgres StatefulSet (infra overlay). sync-wave -1 ensures the namespace exists
+# before the namespaced Secrets/PDB in this base apply (avoids a first-sync
+# "namespaces forte-drop not found" race when the business-apps parent syncs).
+# Prune=false so removing this base never cascade-deletes the namespace (and with
+# it postgres data + the mcp deployment) — matches the earlier decision to keep
+# namespace ownership decoupled from any single workload.
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: forte-drop
+  annotations:
+    argocd.argoproj.io/sync-wave: "-1"
+    argocd.argoproj.io/sync-options: Prune=false
+  labels:
+    app.kubernetes.io/managed-by: argocd
+    app.kubernetes.io/part-of: apps