lisa.wa: Add WA classifiers and id columns in _add_job_info()

This avoids getting ambiguous dataframes for artifacts that don't copy
these data in their own dataframe, like jankbench results.