I have toyed with DHTs successfully over Tor and it works work fine, but that's just peer discovery. If large data transfer doesn't go over Tor, you're still exposing your IP. If large data transfer does go over Tor you're potentially harming the network (even if it's just to onion service) by over using relays (and Tor actively discouraged BT use over the network). There are alternatives like Tribbler, but none popular enough.

Also onion service v3 addresses, which are entire ed25519 public keys, won't fit in an ipv6 address (haven't checked base 32/36/whatever for v2 onion addresses, but surely also won't fit in, say, a /64).

If the tracker itself is also on Tor, you don't need exit nodes, which are probably the most limited resource in the Tor design due to the legal challenges of running an exit node. And while normal Tor clients use a 3-layer design, a reduced number of layers might be good enough for torrent users. If the torrent client participates in the tor network, that would also help.

does tor’s problem with large amounts of data stem from a lack of operators or the nature of tor itself?

I believe a bit of both. I'm no expert on it, so my response here may not be completely accurate.

Tor relays, while more numerous than exit nodes, are still limited (we're discussing a custom onion service communication setup here, not using Tor for general purpose BT which would use exit nodes). Anyone can run one, and please do so if you have good bandwidth each way. Relays are opt in because not only does it open you up to lots of bandwidth usage, but Tor shouldn't default to running a relay for those that may be philosophically opposed. Relays pass all data through anonymously and encrypted, but some still may not feel comfortable doing it. So there are limited operators.

As for the nature of Tor itself, to keep both the client and onion service operator anonymous, they communicate through a rendezvous which means a Tor circuit for each side (each Tor circuit is 3 hops iirc). And geographic distance is intentionally disregarded, so it may traverse the world multiple times over.

I have considered running a private Tor network with directory servers replaced with a DHT, forcing everyone to be a relay, and then essentially running a private BT over that, but it'd need significant adoption to be viable. Also, there may be Tor limitations in huge numbers of relays (and definitely issues with slow relays), I haven't checked. Finally, in addition to traffic analysis attacks that always exist, it is believed iirc lately that federal agencies are sitting on Tor exploits (granted may only be for exits and/or Tor browser).

wow this is pretty ingenious. the only problem is that i think tor has traditionally been a low-bandwidth network?