canbus: document the first-deploy restart trap for rest_command

The alert package looked deployed but the rest_command domain was never
set up — reload_all only reloads already-loaded integrations, so all
four "verified" test fires died with Action-not-found while
check_config said valid. Needs one full core restart on first deploy;
header now says so and gives the verification command.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

canbus/ha/dsi_fault_alert.yaml

@@ -6,6 +6,13 @@
 #
 # Deploy to Pi: /config/packages/dsi_fault_alert.yaml
 # Repo copy:    canbus/ha/dsi_fault_alert.yaml
+#
+# ⚠️ FIRST DEPLOY NEEDS A FULL CORE RESTART. reload_all only re-reads config
+# for integrations that are already set up; rest_command was a brand-new domain
+# on this instance, so until the restart the service didn't exist and every
+# automation fire died with "Action rest_command.discord_server_alerts not
+# found" — while check_config said valid. Verify with:
+#   curl .../api/services | jq '[.[].domain] | index("rest_command")'
 # Secret:       discord_server_alerts_webhook in the Pi's /config/secrets.yaml
 #               (NOT the Gatus webhook — that one posts somewhere Wes doesn't
 #               read; this one is confirmed delivering to the alerts channel)