The combination of `land1` and `land2` are a key to the scores, i.e. if a bullet has six lands, each of the input vectors should have length 36.
max_u(land1, land2, scores, addNA = FALSE)
(numeric) vector with land ids of bullet 1
(numeric) vector with land ids of bullet 2
numeric vector of scores to be summarized into a single number
how are missing values treated? addNA = TRUE leaves missing values, addNA=FALSE imputes with 0.
numeric vector of binary prediction whether two lands are same-source. Vector has the same length as the input vectors.