It's essential that we not take down so many internal DNS zones that if a rack power cycle happens at the wrong time, we have no up-to-date internal DNS zones.
To implement this, the planner needs access to the latest generation number of internal DNS config deployed to each internal DNS zone. We should put this into inventory.
(reference: https://rfd.shared.oxide.computer/rfd/565#_automation)