Distance-from-best threshold, ANDed together

_From @thcrock on July 26, 2017 23:10_

We might want to consider including a second distance from best threshold that the model group can never fall past, even if it gets pretty close at least once, just to weed out some truly bad but very noisy models. I don't want to go too crazy with parameters, but seems like those actually control different things, both of which are meaningful.

_Copied from original issue: dssg/audition#9_

The issue names no file, test, or entry point. Start by locating the existing distance-from-best threshold and model-group configuration, then determine how a second threshold would be combined with it; done would mean both conditions are enforced and covered by appropriate tests.

Written by the indexing model from the issue text.