Burst/cube mode and up-the-ramp sampling for NIR detectors

[oczoske]

ScopeSim (effect ExposureIntegration) currently integrates for the full exposure time (DIT * NDIT) and writes out a single image (usually the average of NDIT subexposures, in ExposureOutput). In burst mode, all NDIT subexposures would be read out individually and saved in a cube, where the third dimension is subexposure number (or time). It would be nice to have this in an effect, with ExposureIntegration being the obvious candidate. This would require to have all subsequent effects be able to read cubes (maybe not that much of a problem as the effects operate on pixels and the dependence on array dimensionality might be easily removed). There is, however, the problem that memory requirements increase quite a bit for large NDIT.
If not in an effect, a loop within readout() might be a solution, as all the subexposures are independent readouts. These readouts would have to be collected in a data cube and saved with a common fits header (augmented by a WCS in the third dimension).

Up-the-ramp sampling is similar, except that the cumulative sums of reads would have to be collected in the cube. Also the readout noise would have to be changed compared to the CDS readout noise used for individual DITs (and currently stored in the irdb). Ideally, detector nonlinearity would be taken into account when computing the cumulative signal.