Want MeanStdScaling that applies to multiple columns

MeanStdScaling computes one set of mean and std params for all the data provided.

There is not a convenient way to compute separate mean and std params for, e.g., a list of columns without looping or creating a separate transform for each.

Maybe a kwarg like :per_col = true but there might be more implication on things under the hood for how this works

Plus one for this, used in GPF on a per slice basis

I think computing mean and std per column should be not only an optional feature but the default one. The current implementation is dangerous because people want to compute statistics per feature (per columns) variable and it is extremely rare to compute single statistics over multiple features.