There are two levels of information available from the GCAM data system:
chunk dependencies, which are available for "free", i.e. with a fast query to
each chunk on the part of chunk_inputs and chunk_outputs,
and detailed information on data object-level dependencies. This function is
used to generate this latter data, i.e. a tibble of chunk-output-precursor information,
which is used by dstrace and various other graphing and diagnostic utilities.
GCAM_DATA_MAPA tibble with columns: name (chunk that produces the data), output (name of the data object), precursors (colon seperated list of precursors), title, units, comments, flags (typically metadata for processing)