The current polynomial fits in the color analysis use numpy.polyfit, which performs a regression on the vertical offsets from the polynomial.  A regression on the orthogonal distance to the fit is preferable (and required for nearly-vertical distributions).  This ticket is to implement an Orthogonal Distance Regression using “odr” uses the scipy.odr module (the numpy.polyfit fit will be retained as it serves as an initial guess for the "odr" fit).

The work here will likely be done as part of the DM-13154 ticket (as plotting limits and refinements may require adjustments as a result of this work).