Currently the ITEM_TABLE class (data_structure package) is defined as
having a list of CLUSTERs, each representing a column of data. This
encoding is not optimal from the point of data representation, since
the most common operation is to add a row, find a row etc. Also, from
the point of view of archetyping, it would be preferable to be able
to define a prototypical row, more or less the way a relational table
is defined.