KoalaBear84/OpenDirectoryDownloader v5.0.0.0 on GitHub

What's new

Faster scans. Scans are quicker, and a scan never fails or stops because of its database.
Web viewer with multiple databases. --serve-db can be given several times and accepts folders. With more than one database the viewer opens with an overview of all of them: sizes, status, live updates and one search over all of them. If the original site cannot be reached the viewer now says so.
One file. The release is a single executable, and --http-cloak also works when you don't start it from its own folder.
Short options can be combined. -qcj is the same as -q -c -j, and values can be given as --threads 10 or --threads=10.
Existing output is kept. An existing URLs file or JSON file is no longer overwritten.

--use-database and --evict-memory are removed: every scan uses the database.
Databases from v4 are NOT compatible

Cut-off downloads: a response that a server cuts off (like lighttpd on large listings) is now continued.
Calibre: no more lost books or random errors, and fractional series numbers work.
Amazon S3: "Slow Down" errors are handled with a wait and fewer threads, and no entries are duplicated.
Fixed listings: SNT index generator directories without files, and Google Drive index detection.
--http-cloak always uses one thread, because it gave errors otherwise.
Browser (Cloudflare): fixed the browser failing to start.
Speedtest: stops when no data is received for 10 seconds and keeps what was measured. It no longer mistakes a differently written URL for a redirect.
Upload of the URLs file: Catbox and 0x0.st are added and tried first, AnonFiles (shut down) is removed, sites that cannot take the file size are skipped, and progress is shown.
Reddit stats: a note about failed directories and a failed upload, and no empty table cells.