Monday 13th August 2012: 11.55am. So, woke up this morning and my "rsync -avc --delete" (synchronise two filing systems exactly, checking each file by bitwise checksum) from the old NTFS external USB drive to new ZFS pool had completed. I had run that after copying over all files to make absolutely, 100% sure that a bit-for-bit copy had successfully worked - and I knew it worked because I had done a ZFS snapshot before the rsync, and it told me not a byte nor a bit had been changed during the checksum comparison. So, I mapped out the old NTFS drive from its mounts and mapped in the ZFS pool via NFS in its place, and voila, everything is now running from the ZFS pool.
Next step is to pull the 2Tb NTFS drive and its 250Gb ext4 overlay temp drive (this was to avoid Linux writing to NTFS on unbacked up data, yeah I know I'm paranoid, but hey overlayfs works very well) and set a 30s idle head park timeout for the 2Tb drive to fix the "unload/load cycle click of death". I then restore the 2Tb drive, use LVM to concatenate it with the 1Tb drive into a fake 3Tb drive, and configure that as a new ZFS pool mirror. Then I can split the mirrors and send each to Canada via different means, and that guards against catastrophe. Hopefully :)