How do you approach testing data extraction and data cleaning processes for a large-scale financial pipeline at Morningstar?