I believe Google can do this already, but it requires each Lemmy instance to tell Google that it’s not the “canonical” version, and which site is. I believe it’s possible for Lemmy and kbin to provide that information already, but in my quick check I don’t see them doing that yet. Will probably need to be a feature request.
Ref: https://developers.google.com/search/docs/crawling-indexing/consolidate-duplicate-urls
How is the import/export business these days?