Decide if it's worth implementing a realistic CCD with the correct pixel geometry

[bregeon]

Right now, images are just 2D matplotlib hexbin histograms, so not even square pixels and these are pretty large.

• the first point is that the above might be enough, as we probably do not need the full 10 um pixel resolution
• the second point it that the smaller the pixels, the larger the images so the heavier the computing aspects

[bregeon]

There are discussions about plugging batoid into imSim, in which case this problem would be solved.

[bregeon]

Note that #41 implements a simple camera geometry overlay on top of a 2D view of the focal plane

[bregeon]

https://github.com/lsst/afw

https://pipelines.lsst.io/modules/lsst.afw.cameraGeom/cameraGeom.html

[bregeon]

Todo

1. create a new function to bin images of the ghosts with adjustable pixel size (10 mu, 100 mu), that may eventually replace the analysis.map_ghosts function
2. create a new analysis.get_ghost_stats function that works from the pixel binned image instead of from raw simulated data

[bregeon]

related to #69