This should be implemented in editors.

It also looks to be nice in exploratory data analysis:

    df = pd.read_csv(f)
    df.columns = map(convert, df.columns, case.snake)