Refactor hardware simulation

[tskisner]

This is to track planned work to change the way that the focalplane hardware is simulated. Here is the current proposal based on various offline conversations:

• When working in cartesian coordinates in the various frames (detector, wafer, tube, telescope), adopt the X/Y <--> Eta/Xi definitions in https://github.com/simonsobs/tod2maps_docs/tree/master/coord_sys

• When doing layouts, work in physical coordinates and layout detectors in wafers, wafers in tubes, and tubes in telescopes in terms of (3D) cartesian coordinates about the nominal (flat) focal surface.

• Separate from the physical layout, define a function that projects detector locations / orientations into spherical coordinates defined with respect to the telescope boresight. In the first instance, we can implement a simple / nominal projection function. In the short / medium term, incorporate another version of this projection function defined by the optics folks and using more detailed information from physical modeling.

This basically involves replacing the quaternions with physical 3D layout coordinates, and then implementing that separate projection function.