COL dataset with thousands of constituents

[mdoering]

The new eXtended COL Checklist on UAT contains nearly 20.000 source datasets in its data and as bibliographic entries in the EML. This causes the creation of thousands of constituent datasets with their own DOI on import of the dwca in the GBIF registry. It also effectively breaks the dataset portal page as the rendering takes minutes - if it ever finishes.

Consider various improvements:

• aggregate sources in the col dwca by publisher used in the xcol config, e.g. plazi, BDJ and a few others
• do not create any DOI for constituents
• wait until we integrate checklistbank.org with the GBIF API. Constituents in the registry might not be needed at all any longer.