Check runner IID on capm3 jobs in order to mitigate Gitlab tags bug

What does this MR do and why?

We are heavily impacted by gitlab-org/gitlab#511377 (closed) for cpam3-virt deployments as we are using CI_PIEPLINE_IID to associate runner to a pipeline. With this bug, all running capm3 pipeline may take any running equinix runner which lead to unexpected behavior.

This change is only a mitigation: if a job starts on a runner which has already been used for a given pipeline, have the job fail this would at least let us have some clean runs, and have clear failures on others (in no way this would be a long-term acceptable solution, of course)