ATM, when drift detection is used and drift is detected (and the stack is set to remediate):
a new run is triggered to remediate.
The original detection has already all the info needed, why not remediate directly? This saves another run, that is largely the same as the original one.