Thanks to changes in restart-server, this is now already happening there.
(The restart-server changes were required to ensure that if the
upgrade failes and one just does
/home/zulip/deployments/next/restart-server to recover, the right
thing happens; so this is the correct resolution to the conflict).