For a delimited importer, it is fine to have multiple rows as the system will merge them, but it is impractical as you have to recreate mappings each time.
You could script the data ingest via the API.
Long term, using a database and a JDBC connector would be the most robust solution.
Modelling the data as RDF and using the RDF importer would also provide a more controlled experience. See the tutorial at:
https://developers.thomsonreuters.com/data-fusion/community-edition/learning?content=11598&type=learning_material_item
See also related question https://community.developers.refinitiv.com/questions/6694/feeding-data-directly-into-datafusion.html
