Added some collapse functions for Python

I added the Python variants of existing SDTL functions col_skew, col_max, col_median, col_min, col_pctile, col_semean, col_sd, col_sum, col_skew, col_var, along with three new functions from Python: col_nomiss (count of non-missing cases in groupby), col_unique (count of unique non-missing cases in groupby), and col_mdev (mean absolute deviation within groupby). I also tried to fix any inconsistencies I found, though I probably didn't catch everything.