Run Brian's script that goes through all experiments and marks runs as redo if their output does not look reasonable.
It could have added functionality like marking as redo for SLO files older than a certain date since we changed the format.