Avoid hitting "Next" downtime by postponing the job later in the night

The frontend SQ analysis frequently fails during the night likely because
it runs soon after SQ is redeployed. Move the update_coverage job 2h
later to avoid that.

.github/workflows/update_coverage.yml

@@ -1,7 +1,7 @@
 name: Update rule coverage
 on:
   schedule:
-    - cron: '17 0 * * *'
+    - cron: '17 2 * * *'
 
 jobs:
   update_coverage: