We could trigger a new run with the same inputs as the failed run. That might be simpler than supporting re-executing the same run.
We could trigger a new run with the same inputs as the failed run. That might be simpler than supporting re-executing the same run.