Because then your browser would itself have to be a torrent client.

The way torrents download is fundamentally different from how a standard http download works, which is why they have a specialist implementation. Browsers dont want to bother bringing a whole load of new code and associated bugs into the browser to do a job which isn’t really connected with the browser’s main responsibility, which is browsing the web.

Just because torrents come from the web shouldn’t make it the browser’s responsibility to deal with them.