Implement scheduled AKS-based functional test workflow and test-failure auto-issue creation #10984
Description
Overview of feature request
Implement a new scheduled test workflow that runs all functional tests against a long-lived Azure Kubernetes Service (AKS) instance. The tests should execute three times per day (matching the current functional test schedule) and will fully replace the currently scheduled functional tests that run in an isolated Kind cluster on the test runner.
Upon any scheduled test failure, an issue should be automatically created in the Radius repository and labeled "test-failure". The new issue should include a direct link to the failed test run, consistent with the current process for auto-created "test-failure" issues on scheduled runs.
The primary goal of this workflow is to identify issues specific to running Radius on a cloud-hosted Kubernetes instance. This addresses scenarios where bugs only manifest on cloud-based clusters, which are not caught when running functional tests solely in Kind on the GitHub Actions runner.
Acceptance criteria
- Scheduled workflow runs functional tests on a long-lived Azure Kubernetes Service (AKS) instance three times per day.
- The AKS test workflow replaces the existing scheduled functional tests running in an isolated Kind cluster/test runner environment.
- Failure in scheduled tests results in automatic creation of "test-failure" labeled issues in the Radius repo.
- Each auto-created issue contains a link to the failed test run, exactly as done in the current Github Actions workflow reporting setup.
- Workflow is documented for maintainability.
- A long-lived AKS cluster is created for running the tests, using the same template used for the Long-Running Tests.
Additional context
Currently, all scheduled functional tests run in Kind clusters within the GitHub workflow runner, which may miss cloud-specific issues. Running these tests in a long-lived AKS environment will help surface bugs that only occur in real Azure Kubernetes environments.