We are using dbt-dataquality and dbt-expectations packages to better our tests. The issue we have been facing is the best way to extract data from the results and manifest json files. We are uploading them into a snowflake stage and then trying to create visualizations out of them. We take those staged json files and using python we create a dataframe which is then fed into our visualization layer. But what we need is to get is the rows that failed the specific test so is that data found inside the json files? What would the best practices be in our case and how should we continue if the data we need is not inside the files?