DHT isn't magic.. every peer builds a cache of other DHT peers over time, but for a new client it'll have no such cache, so falling back to trackers is still important. (IIRC older DHT clients also fell back to a hardcoded "bootstrap.bittorrent.org" or somesuch DNS name)

You might observe when trimming the tracker list off the URL that sometimes you can't find a peer, that's because you've never used the same client to contact what effectively amounts to a subnetwork that has no overlap with any previous subnetwork you leeched from in the past.

Especially if you're always fetching the same content (e.g. Game of Thrones episodes) it'll probably just work, but you may find it breaks when you suddenly decide to grab e.g. an old copy of OS/2 or similar, representing a set of BitTorrent users with entirely different interests.

Nope, it isn't necessary. It just helps with peering, and the actual download will start faster.

Are you sure? I always thought that the peers have to be seeded by the trackers.

How would you as a peer connect to any other peers without some kind of seed?

What IPs do you connect to?

Clients usually have some public, harcoded nodes to bootstrap the search: https://stackoverflow.com/questions/1181301

The trackers are necessary for bootstrapping.