- new transforms for aggregating standard deviation bins with custom bincount
- (prior versions are based on 6 bins for values <2, 2-1, 1-0, 0-1, 1-2, >2)
- binz for onehot-encoded, bzor for ordinal, bzbn for binary
- each accepts parameter bincount for integer number of bins
- bincount defaults to 5
- where if bincount is odd the center bin straddles the mean
- and if bincount is even the center two bins are on different sides of the mean
- also updated inversion transform for standard deviation bins to return bucket midpoint instead of left point