As an engineer working on Pipelines team I want to be confident about the quality of our nightly builds.

Acceptance criteria

1. trigger template "new-index-image" is updated to create necessary pipeline run in addition to mirroring images to quay.io
2. pipeline will install a new cluster (ideally a one-node cluster on PSI) and will destroy it after run
3. pipeline will run in namespace "pipelines-ci"
4. pipeline status is reported back to team using Slack

Notes

if trigger template contains two pipeline runs, they will be created at the same time. Usually mirroring of images is stable and much faster than installing a cluster. In rare cases (e.g. when a new productized image is introduced), mirroring might fail and the other pipeline will fail afterwards (cluster was installed unnecessarily). Is it worth implementing some kind of serialization or is it acceptable behavior?

We need to map Pipelines version (e.g. 1.6.3 or 1.7.1 to the correct operator channel (stable or pipelines-1.7), CLI version (0.21.0 vs 0.23.1) and to omit some test suites when a feature is not available in given version (e.g. Pac, Chains and Hub introduced in 1.7). We can either update UMB service itself or to introduce a simple task/pipeline creating correct pipeline runs.