better detail report on which positioners failed, and which likely "linear" fail

[joesilber]

posparams/plotter.bin_errors() generates a dict classifying positioners by best-fit error over time. For example of how to call load up data and send it to bin_errors(), see plot_posparams script.

Alternately, if you've already run plot_posparams, can just load up the .json file it outputs (which is exactly this result from bin_errors()).

Use this binned data to output some useful reports. In particular:

1. Most recent pass/fail status for each robot, by error threshold.
2. Which robots improve (error diminishes) when binned in "dynamic" mode.

For item (2), this is saying when SCALE_T and SCALE_P are allowed to float. This would indicate likelihood of a potentially usable "linear" type error. To do this classification, need to distinguish "big" changes in SCALE_T or SCALE_P --- a real effect --- from "small" --- just a slightly better fit.