The Ra/dec for a cutouts object represents the coordinates of the middle of the cutout.
But a measurements object will already have some additional offsets given by the measured centroid.
This could include a sub-pixel shift (or more than 1 pixel shift) that needs to be accounted for when setting the coordinates.
These are also the coordinates that get fed into the Object RA/dec.
I'm not even sure anything is propagated into those fields right now. We should explicitly test that the Measurements get an RA/dec and that it is different than the Cutouts pair, and that the derived coordinates get calculated properly (e.g., ecliptic).
The Ra/dec for a cutouts object represents the coordinates of the middle of the cutout. But a measurements object will already have some additional offsets given by the measured centroid. This could include a sub-pixel shift (or more than 1 pixel shift) that needs to be accounted for when setting the coordinates. These are also the coordinates that get fed into the Object RA/dec.
I'm not even sure anything is propagated into those fields right now. We should explicitly test that the Measurements get an RA/dec and that it is different than the Cutouts pair, and that the derived coordinates get calculated properly (e.g., ecliptic).