Brainglobe standardising

[PolarBean]

This PR achieves several things.

• standardises all transformations such that we now assume the provided atlas is oriented the same as the atlasapi. In effect we assume the atlases were retrieved via the atlas api. This addresses [Feature] only work with atlases via atlas api #17
• We begin to introduce semantic versioning with all reoriented transforms now labeled v1.1. (all transforms which lack a version are assumed to be 1.0) Addressing [Feature] Consider moving to semantic versioning #12
• Fixes a small issue with the interpolator where it was using a hidden function which is no longer accessible from scipy
• Copies the atlas api in having one ingestion script per atlas / group of atlases (here we have one script for gubra lsfm and mri)
• Applies black formatting which has caused the diff to be a bit insane
• Saves deformation files to a deformation_fields folder in the .brainglobe directory. This addresses [Feature] Use .brainglobe directory to store deformations #11
• Remove perens STPT as it is not in brainglobe and its not clear its usecase
• Fix a bug which removing perens STPT uncovered
• adds scripts for helping debugging. These plot a point in one volume, and then translate that point to another volume to plot again. If tests are failing this is a sanity check to ensure the points are in the same anatomical location. We also have a tool which transforms each template to every other template (it only uses a selection of the DeMBA volumes so as not to go insane). This is also a good test as many times when I have assumed things were working, this exhaustive test uncovered some strange misalignment.