Shortfall does not work when I(1-px) and/or I(1-py) is missing

[FlorianSeitz]

cognitivemodels/tests/testthat/test_shortfall.R

Line 132 in e4b7196

M1 <- shortfall(choice ~ x1 + px + x2 | y1 + py + y2 , data = dt, asp = ~aspiration, fix = pars)$predict()

As compared to cpt, shortfall only works if all probabilities are specified (i.e., also I(1-px) and I(1-py) must be specified). You might consider making this more flexible?