Log each device when scanned, ignore /dev/nbd*, run blkid with timeout

[tserong]

Problem:
NDM scans every every possible block device under /sys/block, except for /dev/loop. This can cause trouble if there's some weird device that can't be read, e.g. /dev/nbd0 which could be connected when using ceph for external storage.

Solution:

• Ignore /dev/nbd* devices (we know we never want to manage those)
• Run blkid with a timeout, so we don't get stuck if we hit some other weird device that's wedged somehow
• Log each device when scanned so that if we somehow do get stuck even with the above fixes, at least we know what device we got stuck on!

Note: in the original issue I wrote that we should "Limit the disks we scan to only block devices that NDM can reasonably work with", but all I've done here is remove nbd devices. Another approach would be to explicitly only allow sd*, nvme*, dm-*, and ... what else? I don't actually know what a complete list of allowed devices should look like, so I thought it was safer to just eliminate the one that we know we've had trouble with in the past, and not skip anything else. We can always re-visit this later and add more exclusions (or flip it to the style of an "allow list").

Related Issue:
harvester/harvester#9721

Test plan:
TBD

[Vicente-Cheng]

thanks for the enhancement!