Feedback from first run.

Hello - thanks for making this project. Like the simplicity and it mostly worked as advertised. Here are some gotchas I encountered from first run.

I think it'd make sense to add some notes to the documentation so people are more likely to have a good first-run experience:

• Server needs to be ubuntu
• sidekick init disables root ssh. You must use ssh sidekick@ henceforth.
• container should expose port for the sake of health checks since health checks happen from host machine (outside of docker network)
• server should have an /up endpoint also for health checks.

For zero-downtime deploys a few issues:

• The HEALTHCHECK CMD from this example dockerfile caused several seconds of downtime during deploy.
• After removing that HEALTHCHECK CMD... there would still be a few requests which timed out / didn't get a response during a deploy. I suspect this has to do with how nodes are added / removed from Traefik. This is probably why Kamal wrote a custom kamal-proxy... and I wonder if in the long run that might be a good option as it looks pretty nice.