What happened:

Currently the pipeline is failing from time to time on the Step 25: submit-upgrade-test-cloud-build

Based on some logs from the Google console, I've been able to see that there were missing permission for the cleanup, the current roles are based on the deletion of the CRDs but not the whole helm release:
https://console.cloud.google.com/logs/query;cursorTimestamp=2025-08-15T18:12:05.181709256Z;endTime=2025-08-15T18:12:12.289Z;query=resource.labels.container_name%3D%22upgrade-test-controller%22%0Atimestamp%3D%222025-08-15T18:12:05.181709256Z%22%0AinsertId%3D%22s56t9kez37stp8d6%22;startTime=2025-08-15T18:11:53.327Z?project=agones-images&inv=1&invt=Ab5kLQ

What you expected to happen:

The pipeline to work all the time, having the right permission to uninstall the helm release to then delete the namespace without issue

How to reproduce it (as minimally and precisely as possible):

Run any pipeline, a couple of time and it will most probably fails, it's flaky

Anything else we need to know?:

I've been doing a couple of tests with the fix on this PR: #4247

Environment:

• Agones version:
• Kubernetes version (use kubectl version):
• Cloud provider or hardware configuration:
• Install method (yaml/helm):
• Troubleshooting guide log(s):
• Others: