Improve UX for Auroraboot state_dir usage

[jimmykarily]

Originally discussed here: #2522 (comment)

It seems that the way we re-use the state_dir in Auroraboot can lead to confusion. Let's make sure the user is aware of what is being used and when.

[6ixfalls]

In my opinion, auroraboot shouldn't default to reusing the kernel generated initially - there doesn't seem to be any use cases for it. I think a good way to do this is to store the current auroraboot config of the state_dir file and read it when running, and if it's different then regenerate the kernel & whatever else needs to be done. It also doesn't seem like auroraboot retains any other part of the state_dir - it still takes a few minutes after restarting for the server to be up.

[tyzbit]

@6ixfalls I have noticed that using GitHub releases instead of container images makes AuroraBoot much faster and restarts reuse artifacts well, so I suspect that AuroraBoot is slower when using container images because they haveΩ less metadata about the release to make decisions on what cached data can be reused.