CI: improve reliability

[revitteth]

• [x] - only run hive on devel (possibly in future if we can speed this up we could change this back to run on all PRs)
• [x] - stop/remove all containers at the end of every hive run (always())
• [ ] - consider running hive sync tests (quick) on PR just to ensure no simple show-stoppers
• [ ] - add another github runner
• [ ] - add uptime monitoring to runners
• [x] - re-label docker stage to make it clear that the image builds are done on dockerhub
• [ ] - document the CI setup more thoroughly

[J-A-M-E-5]

Hi thought I'd let you know of the following running current dev branch.

Ran a brand new sync without external CI and it synced up to block 15,999,999 processing all blocks. I naively presumed it was then syncing as it kept grabbing all headers up to date with my geth node and with etherscan.

Only when I tried to grab later blocks using RPC did I see that according to the node last block was 15,999,999. It wasn't until I used external CI that I saw it actually grabbed blocks from 16,000,000 onwards.

It looks like a code-path is being missed perhaps just during initial sync from zero? I can send logs if you think they'll help. What I will try and do is once fully synced using external CI I will switch back to using internal to see what happens. Thought I'd comment here on this ticket as I see you're working on the code rather than opening up new one. Cheers.