Miller dunder methods silently discards the phase

[hakonanes]

As reported by @viljarjf in pyxem/diffsims#211, negating a Miller instance silently discards the phase. His example here shows the problem:

from orix.vector import Miller
from orix.crystal_map import Phase

p = Phase(point_group="m-3m")
v = Miller([1, 0, 0], phase=p)

print(v)
print(-v)

>>> Miller (1,), point group m-3m, xyz
[[1 0 0]]
>>> Miller (1,), point group None, xyz
[[-1  0  0]]

[argerlt]

I was going to try addressing this, but to add to the topic,

from orix.vector import Miller, Vector3d
from orix.crystal_map import Phase

p = Phase(point_group="m-3m")
m = Miller([1, 0, 0], phase=p)
v = Vector3d([1,2,3])

print(m)
>>> Miller (1,), point group m-3m, xyz
>>> [[1 0 0]]

print(v)
>>> Vector3d (1,)
>>> [[1 2 3]]

x = v+m
y = m+v

print(x)
>>> Vector3d (1,)
>>> [[2 2 3]]

print(y)
>>> Miller (1,), point group None, xyz
>>> [[2 2 3]]

What's the proper procedure here? I don't think you should be able to combine reciprocal and direct vectors without at least a warning, maybe not at all, but perhaps there is a valid edge case?