Combine morphology extension information with PS rotation information

[Helveg]

In the current temporary implementation if get_region_of_interest of intersectional connectivity strategies we only consider the extension of morphologies as they occur in the MR, and ignore the fact that they could be rotated as dictated by the PS.

I list this as an issue as I'm skipping it while I work to get the proof of concept of parallel connectivity working, but this should obviously be fixed :)

[Helveg]

The difficulty is weighing these 3 things against each other:

• The cost of calculating the rotations the morphologies
• The benefit of having a reduced RoI

Given enough anisotropic rotations, there will be no speedup after going through the effort to calculate rotations: it will practically be the sphere given by the point with max euclidean distance from the origin. So likely, only in very particular cases such as the cerebellar granule cell is the isotropy strong enough to actually benefit from this.

Perhaps a better solution is to allow an isotropic flag on the morphology metadata or cell type for alternative strategies