Why?
The UI requires to have the "full" set of config options for all connectors in place. (in the future even for older versions, when we have multi-version support)
For this we had tests in the UI backend that made us aware when important config "fields" / options got removed, renamed or if new ones were added. Because it was/is a manual process to manage those fields from the frontend/visual/presentation layer.
Since we migrated over into the Connect REST extensions those tests started causing issues when the connector configurations gets changed with PRs.

What?

• Add E2E test/s (running twice a day) that runs those stricter validation checks like we do now for all PRs. Those tests should report those problematic changes (like we did until now with strict tests inside PRs) or general incompatibilities when spinning up the whole stack
• Relax normal REST Extension/s tests that will not easily break PRs but offer good enough stability guarantees
• Fix the generic Connect REST Extension (in module debezium-connect-rest-extension) tests: currently we are using connector artifacts from maven for the Kafka Connect container image, but if for example transforms inside of connector modules are update the tested information can diverge. e.g. count of transforms in the PR is different than the one in the Maven artifact

Sustainability?

• In the future those E2E tests can be used to run other pipelines too, for example: smoke tests for containers, benchmark/performance tests, E2E integration pipeline/s for Debezium source connectors and the JDBC sink connector, etc.

Where?
Should it run on Github Actions and/or the new Testing Farm ™, or setup an OpenShift instance using self-hosted GHA runners with https://github.com/redhat-actions/openshift-actions-runners?