This was done because much of the input data are for the combined unit, so splitting them doesn't add much to the output.
This does not matter for GCAM because they would be merged into the same region anyway.
We do have the spatial data to separate them, however, so we can do it.