There are repos on GitHub that pull the videos and metadata, not sure about posting to Peerhub, though if that's possible to post via an api you could probably script it easily enough. Likely a risk of other "issues" doing so, but I'm sure some datahoarders could chime in.
I'd run this on a VPS if I could do it over the TOR network (don't want to get caught with my CC on the line), but there's the problem of needless duplication if this happens, so it's likely the best if relevant authors do it themselves
I always thought Invidious was downloading the source video from YouTube in the background when given a YouTube video ID and it’s not already downloaded, but I might be wrong.