extended image convolution discontinuities

[JarronL]

There are appear to be significant positional offsets between off-axis PSFs for the NIRCam coronagraphy. This can cause unrealistic discontinuities during the extended image convolution when the PSF spacing isn't very fine.

• Check if the positional offsets are real. Where do they come from? Distortion? SI WFE?
• Implement a solution

Possible solutions:

1. Instead of performing convolutions on a selected region for a given PSF, use RGI to interpolate a new PSF at each pixel position (or finer cluster of pixels).
2. Expand convolution regions to overlap with other PSF masked areas, then average the overlapped regions.