Audio attributes
This merge request addresses issue #110 (closed) , enabling the tracking of extra attributes present in selection tables. In particular, when creating an hdf5 database from a selection table, the user can set include_attrs=True to ensure that any extra attributes in the selection table gets saved along with every audio object. Alternatively, the user can specify which attributes are to be saved using the argument attrs.
As part of implementing this solution, I removed the option to stack several data arrays inside a single base_audio instance, as this was not being used by anyone but added considerable complexity to the code. Unfortunately, this means that the changes in this merge request are not 100% backward compatible, but I highly doubt that anyone will notice, as the functionalities affected were not being used by anyone (to my best knowledge).
See here for an example of how extra attributes are saved to a database file and retrieved afterwards: https://gitlab.meridian.cs.dal.ca/public_projects/ketos/-/blob/audio_attributes/ketos/tests/data_handling/test_database_interface.py#L734