Restarting microshift-observability.service restarts microshift.service and disrupts later tests (like OLM and Gateway API) which look like flaky behavior, but it was unlikely to happen before merging OTEL work.

Also, the test suite fails after the observability suite, if it's before it'll be most likely fine.

1.
2.
3.

Either we rework the microshift.service - microshift-observability.service systemd relation, or we improve our tests (which means something like waiting some time before starting olm / GAPI tests, because at deploy/Pod readiness level they are okay, but they're just not doing very good job)