Performs a full text search on documents in a dataset, then uses those documents to perform an embedding search on the dataset. This is the same as calling /search-documents followed by /search, and using the documents from the first call as a metadata filter.