At present the way we identify sources suitable for aperture measurement is to use sources that the PSF determiner used to determine the PSF. This clumsily entangles measurement with PSF measurement, instead of making measurement self-contained. This leads to awkwardnesses such as the inability to measure aperture correction if we don't fit a PSF model.

The architecture would be much cleaner if we had an efficient way to identify sources that are suitable for measuring aperture correction, and if our measurement task used that.