consistent return type in aggregate family of functions

[gidden]

At the moment, if df.aggregate_region() (and aggregate()) is called with append=False, a pd.DataFrame is returned with a specific, non-conforming pd.MultiIndex. This should be updated to the following:

        kwargs = dict(region=region, variable=variable)
        if append is True:
            self.append(_data, inplace=True, **kwargs)
        else:
            return IamDataFrame(_data, **kwargs)

Then, relevant tests will need to be updated to test for the new return type. An example test is provided in #254 in `test_feature_aggregate.py:

    # # this test should be updated to the below after the return type of
    # # aggregate_region() is updated
    # exp = check_aggregate_df.aggregate_region(
    #     'Primary Energy', region='foo', append=False
    # ).data
    # check_aggregate_df.aggregate_region(
    #     'Primary Energy', region='foo', append=True
    # )
    # obs = check_aggregate_df.filter(region='foo').data
    # assert len(exp) > 0
    # pd.testing.assert_frame_equal(obs.reset_index(drop=True),
    #                               exp.reset_index(drop=True))

[danielhuppmann]

This issue has been resolved in the sequence of PRs #305, #312 and #315.