Use average spectrum to normalise traces for flat fields

We need to normalise traces for the flat-field exposures. We have been doing a "blind find" of traces, extracting spectra and then creating a model image from that, but this doesn't provide any correction in the spectral dimension. PIPE2D-1221 introduces a smoothing of the individual spectra in order to provide correction in the spectral dimension, but it would be better to divide through by an average spectrum.
In order to do that, we'd need to be able to determine an accurate detectorMap for dithered data. This might be accomplished by translating dither values to detectorMap slit offsets, or by a bootstrap-like identification of fibers.