basically what would happen if a function or node should output a pandas dataframe but this dataframe is too large to be stored in memory and therefore you have to process the data lazily