Match pupil to nearest apertures in lenslet array rather than shifting to center

[tepickering]

As the pupil moves around, we need to match each spot to the nearest aperture in the reference rather than move the reference center around. So do aperture match and then find the pupil center to convert positions to pupil coordinates.

[tepickering]

Upon further reflection, I think we're doing this right. The motion of an object in the focal plane registers as wavefront tilt so the spot from a given lenslet moves around as the object moves around. However, in some cases like Binospec where we can't ID a unique lenslet, some significant mis-association is possible.