toplikelihood reparameterization for biphasic poly2

[madison-feshuk]

The poly2 biphasic model update #69 did not adjust the continuous hit calling P3 calculation as seen in #89. Currently the probability of top being above cutoff assumes top occurs at max concentration, which is not always the case: https://github.com/USEPA/CompTox-ToxCast-tcplFit2/blob/master/R/toplikelihood.R#L53

Ex: Fits with poly2 update look better, but this should be a hit. Indicates reparameterization may be needed.