Fix for new paths and restructure to have the from_yml and to_tsv
functions in the same data loading file.
Improved path behavior to resolve any paths to absolutes before
loading.
Ingests all yaml files in the directory, and loads the data for each study.
Currently creates a dataframe for each *observation* in the pool (can be
multiple per study, if a study has multiple analyses for different
independent/dependent vars). This is to follow the tidy data paradigm
of one observation per row.