I would ideally like this to be configurable such that I can apply this logic to pipelines, jobs, tasks, etc. depending on my needs, instead of setting a concourse deployment-wide default_container_grace_time. I would also like the option to turn this feature on and off.

The workflow I have in mind is as follows:

1. I see that a build has failed, possibly a few hours or a day after the fact. This job has been flaky for a while and I haven't been able to get to the bottom of it.
2. This container has not yet been destroyed and I am able to fly intercept in and look around.
3. Once I am done poking around, I press a button that allows the relevant container(s) to be destroyed.