I'd honestly appreciate this enough to pay a subscription for it if there was a policy that meant the subscription fee was explicitly to avoid needing revenue from shady business.

I believe fastmail is like that.

Begging and government grants, like Wikipedia?

There would be drastically less content and the whole operation could be easier, because you probably don't have to worry so much about spam websites. That would keep it low cost already, not single DigitalOcean droplet cheap, but I think it could be sustained from a programmers pocket initially.

Donations/patreon for one, then subscriptions. It should be possible to offer a contentless index for download (I estimate it could be around 50-200GB) and that could be a paid option.