Set a temperature dependent mag limit for imposters

[taldcroft]

Use ideas in imposter stars notebook to set a temperature and box size-dependent mag limit for imposters.

For a given box size, set the mag limit to the magnitude where one expects (statistically) N stars, where N = 2.5 for a 120" box at t_ccd=-10.0. The idea is that statistically on-board you always have imposters fainter than this limit impacting acquisition, and thus they are already accounted for in the acquisition probability model.

With this modification, the proseco probability is only modified if there happen to be brighter-than-normal outliers. This was the original concept of imposters.

This change does significantly impact the estimated p_acq for faint stars, bringing it typically close to the acq prob model.

To do:

• Fix test fails due to catalog changes (which are all minor)

Closes #25

[taldcroft]

Ran validate-2018 notebook with this branch and results are acceptable. The Dragonfly 44 observations still need a hand-edit, but fine-tuning #132 may help.