The idea is to learn about any surprising breaking change and have data to make decision if we should revert or accept the breakage (and ping/help downstream users) -- ideally as an automated CI check on every PR to the release branch.

Some repos we could try setting this up against:

• Update prometheus client golang kubernetes/kubernetes#127941
• Upgrade github.com/prometheus/client_golang to help unblock dependabot. kubernetes-sigs/apiserver-network-proxy#661
• Prometheus
• Mimir

Acceptance Criteria:

• Every release PR (e.g. against release-x.y branch) will have a CI job that creates a draft update PR and proxies their CI status (or similar behaviour).