Support for zip files

We already support gzipped (.gz) and bzipped (.bz2) CSV files. Supporting zipped files (.zip) is a natural next step.

Should we support a zipped collection of files? That is more convenient, but raises questions directory traversal (e.g., do we select all .csv files or only top-level ones?).

This should only require modification to tensor_parser.csv_parser.open_file() unless we add directory logic?