[CLI] allow retries for prove actions

I started a 'prove github' action. I got the keybase.md file up on github but got interrupted so killed the keybase process. When i came back to continue the github proof, I had to start over and replace the existing keybase.md gist (with a new revision).

I would have liked to say "try to verify what I uploaded last time".

While the github experience wasn't so bad to do it a second time, hacker news was a little more annoying. If a site is taking a long time to reflect a change, it's a poor experience to have to leave the process running while I wait instead of either backgrounding it or being able to come back to the same place in the flow later.

I gather from some docs that this process happens anyways? I think it said that Keybase will continue to check proofs over time and notify me if they go away? Exposing this in a more usable way for the CLI might be all I need.

Start by reading the existing CLI prove flow and the documentation describing continued proof checks. Define how a user can return to an interrupted GitHub or other site proof, retry verification without replacing an existing upload, and handle sites that take time to reflect changes; done should include a documented, usable retry or resume flow.

Written by the indexing model from the issue text.