Currently, we decommission expunged disks during reconfigurator execution. This expungement step is tied to a previous step in the executor that ensures that any expunged disks no longer exist on the given sled.
Unfortunately, this coupling prevents us from making all execution steps independent and skippable as a solution to #6999.
We can resolve this issue by decommissioning disks in the planner, rather than in the executor as we do with sled decommissioning