Well, from a software engineering perspective, you can easily have a bug that is fired off from the throttling method if they added that method to a new procedure, but did not account for all outcomes.
In other words, it could be throttling but not saving it to the location. And the new procedure might be regarding temporary flctuations of the grid, not the local charging equipment. So there could be separate code to detect faulty equipment, and a new piece of code to detect grid fluctuations using the same detection method, but the grid method might not have the outcome of saving it to the location....
Anyway, this is all speculation. However, I still belive it has nothing to do with scheduling of charging or any kind of "smart" charging. Just unintended consequences of reusing peices of old code in new funtionality. It might already be fixed.