Multi-index columns are not produced properly in the dataframe_to_rows function

There is a bug introduced when the expand_index function was generalised and added to dataframe_to_rows (https://foss.heptapod.net/openpyxl/openpyxl/-/blob/branch/3.0/openpyxl/utils/dataframe.py#L64)

A Multi-index column of this shape:

col_name1     col_name2                   col_name3
col_name1.1   col_name2.1   col_name2.2   col_name3.1    col_name3.2

Is transformed into

col_name1   
col_name1.1   col_name2.1   col_name2.2   col_name3.1    col_name3.2

missing any of the subsequent column names in the first level FYI @charlie.clark