I'd mostly focus on scrub, but I'll also recommend that you keep an eye on your unallocated blocks, running out of them can get you into out-of-space situations that are non-trivial to fix. My general recommendation is about 5G per device, that should give it enough breathing room that it won't -ENOSPC on most workloads. Also, please note that unallocated space is a subset of free space, that is, all unallocated space is free space but the inverse isn't true.

Getting more unallocated space is as easy as running a balance with a filter, say "btrfs balance start -dusage=10 /mountpoint". Just don't balance metadata unless you want to convert it.

Regarding defrag, I still defrag databases, system journal files, etc, even on SSDs. Those workloads tend to cause a lot of fragmentation that can impact your performance (try reading your journalctl logs before and after a defrag, as an example).