Need a query to delete duplicate records in ETL import tables

1The most recent FamilySearch imported database contains 205M records, 85M of which are duplicates. There are two tables involved in the ETL process, ETL.importRecords and ETL.importNames. The tables can be joined through a column named "RowNBR". The duplicate records occur throughout the table and for anyset of duplicate records, the record with the highest RowNBR should be retained and the record(s) with lower RowNbrs deleted. When records are deleted from ETL.importRecords, the corresponding records with the same RowNbrs in the ETL.importNames table should also be deleted. The duplicates may be identified by the RecordOriginalText column in the ETL.importRecords table. RecordOriginalText is an varchar(max) column, and entries look like
"000000000_004949962_00177_00083_00001109434222". The entries represent the FamilySearch recordID.