We know that we're not yet able to do everything we want. New features include, but are not limited to:

• Direct image scaling and subtraction
• Leveraging information in both m=+1 and m=-1 orders (for coarse grating)
• Using the wcs, or a fit to the wcs (when possible) to get an exact centroid for the main star

We probably also want to improve some of the priors once we know how things are working, which will improve fit robustness, and perhaps also allow finer sampling of some of the pre-computed grids which are fit against.

Likely will need to be split once we know more about this work.