Recover disk space on staging

Our regular updates fill up the staging servers' disks rather quickly, and we have to clean up our staging system too often.

Since we send logs to a central server, we could limit the journald logs more aggressively (say, only keep a week of logs on the machines themselves).

Also, while using swap to have a way of saving our machines from falling over in an out-of-memory situation - get alerts & jump to fix the problem before the memory is exhausted - is nice in theory, this never worked in the past (i.e. when nix upgrades exhausted the memory) and also, staging is just not that important and I would rather have it fall over sooner than later. (i.e. fail fast and hard to make problems very visible)