pub const DOWNLOAD_CONCURRENT_THRESHOLD: usize = _; // 10_485_760usizeExpand description
Routing threshold for downloads: below this, Client::download uses a
plain single-connection stream; at or above it, downloads go through
the concurrent/pipelined engine (Y worker connections, each X requests
deep). Independent of BIG_FILE_THRESHOLD on purpose - that constant
is upload’s protocol-mandated saveBigFilePart boundary, and reusing it
here was coincidental, not protocol-driven. Same value today, but free
to diverge since nothing in the download path actually depends on it
matching the upload number.