88 flipped hitc

[brown-jason]

Cutoff now takes into sign(top) into account.

[madison-feshuk]

See https://github.com/USEPA/ToxCast-Internal/tree/main/analysis/tcplfit2_hitc_toplikelihood
for a standalone analysis reviewing this issue and fix (Rmd and html). Slides featuring the reprocessing results on the ToxCast database, which were presented to HTTr/HTPP/ToxCast teams on 6/10, are also provided.

ToxCast team approves this toplikelihood fix for its invitrodb v4.2 release. In the release note, an known issue will be listed related to poly2 reparameterization which may be skewing toplikelihood calculation #90

[leverett-epa]

Code changes make sense for my understanding of the issue, and testing appears to have demonstrated the validity of this approach