add_camera_rigs: refactor and cleanup

- Fix widgets’ names: they were hardcoded and didn’t follow the
  preferences, leading to crashes.
- The UI was put back into the Item category, instead of Create,
  because it is not related to object creation.
- Fix some strange topology in two widget shapes.
- UI and operators use a new poll method, so that they work when
  either the rig or the camera is selected.
- The composition guides UI was converted to a panel, so that they may
  be drag-selected.
- Marker binding and DOF object operators were converted to the
  `bpy.data` API, making them simpler.
- Bones were moved around so that they are more similar between rigs.
  - They were scaled down to be 1 unit long, a simpler length — for
    instance, widgets are the same size as modeled. Widgets were
    scaled up to compensate.
  - The camera and aim bones were placed at 1.7 unit high, to be
    approximately at a standing human’s eyes’ height if the scene is
    in meters.
- Much of the rig generation was refactored to deduplicate code
  between the two rig types.
- Automatic renaming to `.000` was removed, since Blender already
  handles duplicate names.
- Widget prefix and collection were renamed to `WGT-` and `Widgets`
  respectively. This is to be closer to Rigify, hopefully unifying
  them.
- The GPL license header was added to every file.
- Some cleanup was done to better respect Python’s PEP 8.

Reviewed By: Wayne Dixon

Differential Revision: https://developer.blender.org/D6543