FreeNAS Code

- Only restart smartd when needed.
- Fix notifier in a different way. Creating gpt partition changes
the topology so we need to invalidate cache there.

make the disk replacement works if there is more than one disk unavail

Try to cache confxml but not too aggressively.

half-way solution for zfs disk replacement

introduce a Pool object while keeping the legacy API for now

middleware method to parse a given pool name

Don't do down-then-up cycle when starting if IPv6 is already
enabled.

move some classes/method related to zfs for future use in replacement;
a lot of work is still required for a better API for a more pythonic
way of retrieving the state of the pool

Merge in a few revisions from iXsystems' repository:

- Use -f when attaching zpool vdev. In the case the device is not totally
erased, we would not be able to attach them without this change.
- Eliminate unneeded multiple zfs get runs. This is mostly a rewrite of
zfs_get_options by adopting get_zfs_attributes's ideas that only runs
zfs(8) once and get all attributes in a batch.
- Remove the unused [sg]et_zfs_attribute[s].

make sure newfs for single disk doesnt hit the .wait() pipe limit