There are tons of e.g. 8TB SSDs or 20TB HDDs, etc. but they don’t always have exactly the same size when looked at available bytes.
So if say a few years from now such a drive dies, and needs to be replaced, there may no longer be the exact same model on the market, instead there might be other 8TB or 20TB devices around, which just happen to have slightly different sizes, which might be a little bit smaller due to implementation details.
Is there a way, and if so what’s the recommended practice, to keep a safety margin, such that one doesn’t end up getting stuck with a dead pool because one can’t find the properly sized device? Or is the solution to simply replace with a significantly larger drive e.g. a 24TB drive, just to avoid the scenario?
This existed in FreeNAS, in CORE, and in early versions of SCALE with the 2 GB swap partition–its real purpose wasn’t to add swap to the system, but to allow for such drive capacity differences. With SCALE 24.04, it went away, causing some trouble of exactly the sort you’re describing. In either 24.10 or 25.04 (don’t remember which), something similar was returned–swap partitions aren’t being created, but the data partition is about 2 GB less than the size of the drive, for the same reason.
Cool. Thanks. That’s good to know. Since I’m setting up a new system, I should be all set, in that case! 
No system is ready until you run a memtest bro.
Done that pretty much first thing, because if the memory were bad, I’d want it to be exchanged under warranty…