The download command will be CSV/TSV aware in that:
you can download the first n --sample records
and for servers that support Range Requests, be able to specify an offset (in MB) before downloading the following n --sample records, sniffing the head of the CSV first to determine the CSV schema
take advantage of the same Range Request server support typically used for resumable downloads, to get new records that have been appended to a remote CSV file, downloading only the new records and appending it to the local copy.
be able to download multiple files (using http2 and brotli/gzip/deflate compression if the server supports it) using a glob pattern if a web server supports directory listing
The
download
command will be CSV/TSV aware in that: