The prototyping and understanding the cost-benefit analysis of the "interpolation" is being carried out in the Jupyter Notebook here: https://github.com/lsst-dm/GAaP_analysis/blob/master/GAaP_interpolation.ipynb

Looks good. I made some mostly minor suggestions to enlarge & document the first plot. I am curious to see the comparison for more galaxy-like images.

Summary: The prototype done in the Python notebook linked demonstrates that sinc coefficients are not required to compute fluxes with Gaussian apertures for the aperture sizes we would consider in practice. Thus, for GAaP, it will suffice to evaluate the Gaussian profile at pixel centers.