[wip] When provided external data, OD calibration curve bifurcates

[CamDavidsonPilon]

Original report: https://forum.pioreactor.com/t/manual-od-values-gets-registered-higher-in-calibration/273/5

A few curiosities first:

1. Why were the data manual data points nearly 1-1 with the inferred data points (along the x-axis). This suggests that the dilution amount was >= 5ml, since 10 + 5 + 5 >= 18, so the algorithm would want the user to replace every 1 dilution (instead of 2, if 4ml dilutions were used). However, this contradicts what the user said, and it would be odd for the user to not notice that one (two) of their Pioreactor kept asking for cycles faster than the others.
2. The chart is showing "strange" x-values. Part of this is that plottext, our plotting lib, has terrible defaults fixed in next release. It's hard to get good data from a tool like web-chart-digitizer when your scales are not equidistant!