reduceProfile fits fiber profiles simultaneously to all fibers for multiple exposures. In order for this to be a linear fit (and therefore relatively fast compared to a non-linear fit), we need to have an independent measure of the amplitudes of each fiber in each of the exposures. This is done iteratively: we extract spectra using a guess profile, and use that in fitting the profiles, and then repeat. In the end, we have the fiber profiles and the amplitudes (from the spectra). The profiles are the main result that we want, but the amplitudes have useful information on the quality of the fit. We should persist these and evaluate them to ensure they match our expectations given which fibers have been hidden, etc.