Recently, we have had several regressions that could have been captured by OCP tests during pr-checks. However, the truth is that the ocp-smoke-tests label is not currently used for pr-checks because the selection of tests is not optimal - in its current form, a ocp-smoke-tests pr-check would ingest all existing mutli sources and consequently take almost as long to run as a full-smoke run. 

We should improve test selection for the ocp-smoke-tests pipeline so that the pr-check is fast enough, stable, and provides good test coverage for OCP-related changes, and then encourage the dev team to start using it