Automate refresh of database to staging
We used to do this on a daily basis, see the thread in chat, but it was done with a one-off script. @northrup points out that "when we refactored the fleet we didn't restore that function to a worker as we split and separated them."
As far as I understand, the reason to do this is not because we need the data per sé, but because various tests and deploys with reverts on staging can end up leading to hysteresis: the database ends up in a state that is no longer representative of the state from which it will be expected to start upon the next deploy.