After an upgrade to systemd 247.3-6~bpo10+1 from buster-backports, I can no longer reproduce the problem in 15+ minutes of trying, when it always took less than 1 minute before. I think it's fixed. Sorry for the noise.
Starting at: https://bugzilla.redhat.com/show_bug.cgi?id=1845456 systemctl try-restart command hangs indefinitely while being executed during a yum update ... I think the fix might have been: https://github.com/systemd/systemd/pull/13124 core: coldplug possible nop_job That's not in: https://github.com/systemd/systemd/blob/v243/src/core/unit.c#L3925 ... but is at: https://github.com/systemd/systemd/blob/v244/src/core/unit.c#L3894
