New system for ultrastructural features

[schlegelp]

connectors are currently more-or-less hard-coded into navis. In the future we might need to be more flexible for when people want to associate other kinds of meta data with their neurons. Examples for this would be mitochondria or dense-core vesicle.

What I have in mind is a class system where each class implements a couple relevant methods for e.g.

• data validation (e.g. requiring x/y/z/type columns for connectors)
• plotting (in 2d, 3d & for plotly, matplotlib, Vispy)
• summary statistics (e.g. "n_connectors" for connector tables)
• ...

Leaving this issue as reminder to self and in case others have thoughts/suggestions.