Once you run explore_data_descriptions, you should be able to use the valid sample ids and dates for the later PipelineSampleGetter easily.
Once you run explore_pipeline_sample_getter you should be able to use the output to define a new PipelineSampleGetter which skips the failed sample ids and dates.