The way to take advantage of the bandwidth of multiple storage devices is to distribute concurrent writes across them, rather than forcing everything to commit sequentially using contended locks or rollbacks.
The DB (or any application) should not have any need to know what devices are underneath its mount point. If you’re striping across disks, that’s a device (or filesystem, for ZFS) level implementation.
Using a monotonically increasing PK would cause pages in the index to be allocated and filled sequentially, increasing throughput.
Using random UUIDs would lead to page-splitting and partially-filled pages everywhere, negatively impacting performance and size-on-disk.