Managing pixel coordinates within Detections

[MatthewWhiting]

A perennial problem I have is keeping track correctly of what pixel coordinate system I'm working in - is it the global image coordinates, the subsection coordinates, or some smaller cutout coordinates?

We need a new design to better manage this. We could have a set of different coordinate systems, that each come with a subsection definition and an identifier, then a switching command that chooses a particular system, and adds any appropriate offsets given the current identifier and the desired target identifier.