Not necessarily, it's possible that you could only use a Threads account from a Threads app. But hypothetically anyone on a federated instance could interact with a Threads user from a non-threads instance/app.

I would like some kind of content-warning type situation, but where I can define what kind of content I want to restrict. And then, somehow, instances have tags associated which then apply to all of their users so that I can have a warning that users of a given instance have a reputation for sharing content that I would like a warning about first.

I would still want the content to be posted, but I would prefer that it requires me to actively unhide the content and default to always seeing it.