feat: impl new supervisor traits k8s

[DavSanchez]

Implements the new traits defined on #2048 for the Kubernetes running mode of AC. The implementation actually stops and destroys the running supervisor and starts a new supervisor under the apply method (making it a bit of a misnomer). This is mainly a matter of thread management ergonomics, but this could be reworked if desired (this PR or later iterations) by adding some data sharing.

Checklist

• Provided a meaningful title following conventional commit style.
• Included a detailed description for the Pull Request.
• Documentation under docs is aligned with the change.
• Follows guidelines for Pull Requests in CONTRIBUTING.md.
• Follows log level guidelines.

[danielorihuela]

Cool! I think it reads better, and we can improve it a little bit more I think. What about adding the logs directly to the stop_blocking function?

/// It sends a stop signal and waits until the thread handle is joined.
pub fn stop_blocking(self) -> Result<T, ThreadContextStopperError> {
    ...
    self.join_thread()
        .inspect(|_| debug!("Thread {thread_name} stopped"))
        .inspect_err(|e| error!("Error stopping thread {thread_name}: {e}"))
}

fn stop(self) -> Result<(), ThreadContextStopperError> {
    let stopped_threads = self
        .into_iter()
        .map(|ctx| ctx.stop_blocking())
        .collect::<Vec<_>>();

    stopped_threads
        .into_iter()
        // Return first err
        .try_for_each(identity)
}

Also, using some variables to differentiate steps might help. But this is purely subjective.

[DavSanchez]

I was initially against it to not change much of the existing implementation, but I can see the advantage and finally moved the logs as you suggested, thanks!

[danielorihuela]

Maybe you can have a separate PR with these improvements if you want. I think it's more work.

[DavSanchez]

The logs were already moved to the stop_blocking and there's a bit of comment to indicate what exactly identity is for. Let me know if you'd like something else!

[gsanchezgavier]

could you include some PR description ?

[DavSanchez]

could you include some PR description ?

Certainly! 🚀

[danielorihuela]

I like the new approach. We avoid having a special "try_build_new_supervisor" or something like that as before. Cool!!!

[gsanchezgavier]

it is an arc right? if so perhaps i would remove this comment that brings some uncertainty

[DavSanchez]

Yeah it's an arc, I'll delete the comment or specify we clone the reference

[DavSanchez]

Remove misleading comment c976700

[gsanchezgavier]

Cool thanks!, I just left a question

[gsanchezgavier]

🤔 is it ok to stop , wouldn't that leak threads ?

[DavSanchez]

Yeah the threads are all stopped in the previous binding let stopped_threads, this only selects the first error among the results (a behavior similar to the one we had before, perhaps it's time to aggregate the errors and return them all instead?)

[gsanchezgavier]

ohh i just see these are the returned errors of the joins !

perhaps it's time to aggregate the errors and return them all instead?

It looks enough as it is now for me 👍

[gsanchezgavier]

🚀

[gsanchezgavier]

ohh i just see these are the returned errors of the joins !

perhaps it's time to aggregate the errors and return them all instead?

It looks enough as it is now for me 👍