Definitions related to Symmetry versus Point Group

[argerlt]

I think orix would benefit from some stricter terminology constraints in the Symmetry module.

In #563 , we are talking about adding a PointGroups class which is a collection of Symmetry lists, but Symmetry has a docstring that starts with:

The set of rotations comprising a point group.

similarly, osm.get_point_group returns a Symmetry object, which is confusing if we add a PointGroups class.

In my mind, Symmetry is just the concept of rotational and/or mirror operations, so for example a five-fold symmetry is allowed. A Point Group is a more specific term, referring just to symmetries possible in a crystal. This distinction is small peanuts now, but important if we start talking about Specimen symmetry in the future (which we should).

One possible implementation is:

class PointGroup(Symmetry)
--> just symmetry, but with Schoenflies names and required to be crystallographicall possible.

class PointGroupList(PointGroup)
--> mimics the concept of PhaseList. Includes functions to return interesting lists described in #563, like:
        - proper
        - laue
        - proper_permutations

[CSSFrancis]

+1 for better distinctions.

There are lots of symmetries that aren't crystallographic. Things like quasi crystals and amorhpous materials. In my phD I dealt quite a bit with non crystallographic symetries and it would have been nice to have a better way to define them.

A symmetry should be something we could apply to an image or a 3D object. Even taking a 3D model, applying a symmetry and measuring how far from symmetrical it is.