@vaceksimon / @mhajas - this is a PR which avoids nip.io in the reported places. Please review when you have the time and merge. Thanks!

For the cookie provider test, 0.0.0.0 is not considered a secure context as long as no https is used, so IMHO it is equivalent to .nip.io and it is not considered a localhost address. I see this regularly when opening that URL in the browse locally and the APIs to copy to the clipboard are regularly not working.

For the reverse proxy it doesn't matter as there are both http and https test. And using 0.0.0.0 didn't work here, as the client will be connecting via the public address, and then the tests will fail.

@jonkoops - AFAIK you have been involved in the discussion about secure vs. non-secure contexts. Maybe you can help to assess this change.

0.0.0.0 is not considered a secure context (see specification), so I agree with @ahus1's assessment that this is likely equivalent. We only need to be careful with tests that might do cross-origin requests (need to be on different domains), as those might result in a different result for setting cookies, but I see no such changes here at this time.

Note that 127.0.0.1 (previously proxy.kc.127.0.0.1.nip.io) is considered a secure context, but I am not quite sure if the context for the reverse proxy was ever used securely anywhere that impacts user-agents.