...
If it's only going to be subjected to its highest temperatures during the occasional replication or scrub, is it really that dangerous to the drive within? Wouldn't it serve as a sort of make-shift "stress test" on your drive, in the same spirit as stress testing and "cooking" a new CPU to make sure that if it's going to fail, you'll catch it earlier than later?
In theory, except in practice a scrub might last 5 HOURS on large disks, over less than speedy USB interface.
Let's be clear, many external enclosures use USB, which as a temporary connection is fine. But, even USB 3 speeds, 5Gps, is less than ideal when compared to SATA III. I am NOT talking about the different in USB 3, 5Gbps verses SATA III, 6Gbps. No, what I mean is that the hard drives native interface is almost certainly SATA, not USB. So, the old USB Block Transfer protocol is slower, causing any long task, (like a ZFS scrub), to be even longer.
That said, USB UASP, (USB Attached SCSI Protocol), does improve things greatly. But, both sides of the connect need to support UASP, which includes the USB drive enclosure.
Having a hard drive heat up very high every backup, (my backup disks rotate every month or so), first due to a ZFS scrub. Then second due to the actual data transfer, is not something I want my backup disks exposed to.