Replace zfs-related subprocess calls by library calls

[Defenso-QTH]

This is a sub-item of #95.

In a profiling iocage list run I made of 0.795 seconds, 0.204 seconds were spent in zfs.py's run function, which is just a subprocess wrapper to call zfs(8) or zpool(8):

iocage/iocage_lib/zfs.py

Lines 8 to 23 in 6518661

	def run(command, **kwargs):
	kwargs.setdefault('stdout', subprocess.PIPE)
	kwargs.setdefault('stderr', subprocess.PIPE)
	kwargs.setdefault('encoding', 'utf8')
	check = kwargs.pop('check', True)
	proc = subprocess.Popen(command, **kwargs)
	stdout, stderr = proc.communicate()
	cp = subprocess.CompletedProcess(
	command, proc.returncode, stdout=stdout, stderr=stderr
	)
	if check:
	try:
	cp.check_returncode()
	except subprocess.CalledProcessError:
	raise ZFSException(cp.returncode, cp.stderr)
	return cp

We could reduce this time by using libzfs, avoiding the process creation and communication overhead.

[dgeo]

current libzfs port fails to build here (1.1.2023020700), not sure how well it is maintained for FreeBSD…
see https://www.freshports.org/devel/py-libzfs/

[Defenso-QTH]

I wonder why there is no py311 table on that page.
It is already used in the tests:

iocage/tests/data_classes.py

Line 10 in fbb0003

import libzfs

And that passed CI with installing it through pkg so it's available.

[dgeo]

my mistake: https://www.freshports.org/filesystems/py-libzfs/

Let's use it :)

[dgeo]

make this for 'future', will move to 'next' if we can test this thoroughtly in time