Add intrapixel sensitivity map downsampling

[benjaminpope]

In this paper there are intrapixel sensitivity maps measured for H2RG detectors in the lab, at 650-1550 nm. This isn't the wavelength range we want, and there are significant inter-pixel variations too (which we have solved in Desdoigts et al 2023a), but the intra-pixel stuff is important too if we really want to nail high angular resolution, high contrast imaging like with AMI.

We could add a version of downsample, whether as a Detector layer or a util, that instead of taking the mean of the array along the chosen axes, dots with an n by n array containing the mean intrapixel sensitivity map.

We could publish a fit to JWST data with mean intrapixel sensitivity maps inferred from data at each wavelength.

[LouisDesdoigts]

Oh yeah this is a really cool idea - definitely something we build and do in due course