TPA-RFC-20: plan a bunch of bullseye upgrades

in a recent bullseye upgrade (#40654 (closed)), I found the upgrade took 40 minutes. i have therefore reasons that, in a work party, we could probably upgrade a significant chunk of our fleet.

so the task here is to:

• look at our inventory, and figure out which servers should be fairly easy to upgrade in batches
• create an issue per machine to be upgraded? we created issues for batches (#40690 (closed), #40692 (closed), and other related tickets here)
• then schedule a work party to massively upgrade those, all at once
• send an announcement to affected parties that their servers will be upgraded (TPA-RFC-20 sent to tor-project and forum
• do the actual upgrade see related tickets for progress

part of this work is to consider whether it's worth writing a batch script (fabric? shell? mitogen prototype?) to do parts of the upgrade...