Change confidence calculations to represent true positive rate

Our current confidence ratings are pretty useless. What we REALLY want to output is the percentage chance that the hit we have observed is a true positive.

There are three factors that can affect the likelihood of a false positive: depth of the prng searched (greater depths increase false positive rates), bounding (tight bounding increases false positive rates), and count of input numbers (small numbers of inputs increases false positive rates).

Let's do the (complex) math on that calculation and inform the user what their true positive rate is. Also, we can tell the user how to increase the rate by adding more inputs. (Reducing depth and changing the bounds aren't something you as an attacker can just do)