Misorientation.reduce() tosses proper/improper information

[argerlt]

When working on #669, I noticed the following operation wipes out all proper/improper information:

import numpy as np
from scipy.stats import norm
import orix.quaternion as oqu
import orix.quaternion.symmetry as osm

m_as_is = oqu.Misorientation.random(1000,symmetry=(osm.C2v, osm.Oh))
m_as_is.improper = np.arange(0,1000)%2>0
m_reduced = m_as_is.reduce()

print(np.mean(m_as_is.improper))
print(np.mean(m_reduced.improper))

Do we really want to do that? We definitely want to include inversions and rotoinversions when defining the bounds of the fundamental zone, but I think wiping the inversion from the reduced result might be problematic.

If the answer is a definite "Yes this is should be fixed", I can fix it as part of #669, otherwise we can push #669 and discuss this more if needed.