Add option to use tweaked detectorMap to define fiberProfiles

Looking at the discrepancies between the detectorMap and the fiberProfile centres shows that the Chebyshev fit to the x-centres is pretty good, but not perfect.  Using the detectorMap (probably with an affine distortion) would be better.  The figure shows every fifth fibre, and clearly some of the fits are a bit off (much worse in the dense-pack region).  Using the black spots will help here, of course, but a global distortion would probably be better.