Investigate wavelength solution quality

We're seeing wavelength solutions with an RMS of the order of pixels, which is at least an order of magnitude too high. Also, the arc lines from the different fibers don't line up after wavelength calibration. Investigate and fix.

rhl recommends looking at visit=5830 from October 2017, for which he has a plot indicating a good wavelength solution. We can also look at the sim data.