Right now you can't extract table metadata without using Pandas. In order to at least get something out, we could implement a very lazy loader that only extracts data files lists and factor values to get some summary data from the tables without having to do a full load into memory of the experimental graphs.