app: drop dead whenever cron line; restore --atomic deploy

The app image has no cron daemon, so whenever --update-crontab was a no-op —
the daily sitemap:refresh it installed never ran. The sitemap must live in the
serving pod (public_file_server serves /sitemap.xml), so a CronJob is not
suitable; the on-boot sitemap:refresh is the correct pattern and is kept.
Restore --atomic now that the deploy is healthy (was --wait for inspection).

Author: cpcwood

.docker/scripts/startup-app

@@ -14,6 +14,5 @@ else
 fi
 
 bundle exec rails sitemap:refresh:no_ping
-bundle exec rails whenever --update-crontab
 
 exec bundle exec rails server -b 0.0.0.0 -p 5000

.github/workflows/main.yml

@@ -552,6 +552,6 @@ jobs:
         run: |
           helm upgrade --install home-server ./infrastructure/helm \
             --namespace home-server-production \
-            --wait --timeout 5m \
+            --atomic --timeout 5m \
             --set app.image.tag=${{ github.sha }} \
             --set worker.image.tag=${{ github.sha }}