EDIT

Common base classes already exist. Further work might be:

• use the same calls for "aggregateWithinTaxon", "aggregateToHigherRank"
• use TaxonNodeFilter in structured description agg. similar to distribution aggr.
• use memory handling like in distribution agg.
• merge source handling
• do not create new objects if object exists already from previous aggregation
• discuss dataset handling for distribution aggregation (DONE for DescriptionAggregation)
• tests for includeUnpublished
• tests for sourceType
• fully delete empty aggregations
• ...