extend transform_columns to support groupby operations

[samukweku]

Brief Description

extending transform_columns to groupby operations could make some wrangling steps a bit cleaner/easier? StackOverflow

Example API

# transform only one column, while creating a new column name for it.
df.transform_columns(column_names=['col1'], function=np.abs, new_column_names=['col1_abs'], by = ...)