The precision-recall, gain and lift examples move to an outcome with a scarce positive class, the case those charts are documented for. They previously ran on an outcome that was 94 per cent positive, so the gain curve sat on the diagonal and lift hovered at one.
The calibration example fits a model rather than plotting a hand-written score, and the power-curve article names the effect the shipped simulation actually covers, with its no-simr branch derived from that simulation so the two agree by construction.
vif_plot() restricts its scale to the severity levels present, removing an empty key entry from the rendered figure.
R CMD check --as-cran: 0 errors, 0 warnings. Not yet submitted to CRAN.