Identify appropriate return type for empty aggregation results with UDF aggregations

In #8274 the output of aggregations returning empty data sets was changed so that the output type of nested columns would match the expected type even when it is empty. However, this led to a separate edge case, which is determining what data type to return when performing a UDF aggregation that returns an empty column. A numba-compiled Python UDF provides enough information to infer the output type, but we need to assess the extent to which UDFs returning nested are/should be supported in cudf Python to determine whether/how to pass this information down to `libcudf` to ensure that the appropriate types are returned. As of #8274 such aggregations simply mimic the input data type for the returned column.