fix(k8s): handle existing workspace on pod restart

The clone init container was failing with 'destination path already exists'
when the pod restarted. EmptyDir volumes persist across container restarts
but init containers run again.

Now checks if workspace exists:
- If .git directory exists: fetch and reset to latest master
- If not: clean and clone fresh

This fixes the CrashLoopBackOff caused by failed clone attempts.

k8s/deployment.yaml

@@ -22,10 +22,20 @@ spec:
         - |
           set -e
           apk add --no-cache git
-          git clone --depth 1 --branch master \
-            http://gitea.gitea.svc.cluster.local:3000/unchained/manoon-headless.git \
-            /workspace
-          echo "Clone complete."
+          # Clean workspace if it exists (handles pod restarts)
+          if [ -d /workspace/.git ]; then
+            echo "Workspace exists, pulling latest changes..."
+            cd /workspace
+            git fetch origin master
+            git reset --hard origin/master
+          else
+            echo "Cloning fresh repository..."
+            rm -rf /workspace/*
+            git clone --depth 1 --branch master \
+              http://gitea.gitea.svc.cluster.local:3000/unchained/manoon-headless.git \
+              /workspace
+          fi
+          echo "Clone/update complete."
         volumeMounts:
         - name: workspace
           mountPath: /workspace