CSV loads directly into the database would entail a LOT more work than processing a tab-delimited file and creating the SQL statements, even if there was an easy way to get the CSV file through the firewalls to the database server.
Also, there are fields in the database that you wouldn’t be able to fill out, because they’re used internally for sequencing and are dynamically generated.
I also see separate tab-delimited files for each section as more work for everybody than a combined file for the entire event, because I’d either have to deal with a variable number of files to import and what happens if a section number is skipped or duplicated, or you’d have to upload each one separately, at which point each section would be assigned its own USCF event ID, also dynamically generated by the database server, which would mess up searches on MSA.
I need to acquiesce on the csv files for PostgreSQL. And, I see your point with a file for each section. There would have to be coordination of the torunament ID and, as you said, a section could be totally skipped or not even submitted.