Suggestion: ability to specify a node range in flux jobs

[kkier]

Scenario:
As a sysadmin, I'm planning to do maintenance on an arbitrary set of nodes. After draining those nodes, I'd like an easy way to return a list of jobs running on them so I can contact the job owners/see when they'll expire/kill them outright/whatever. Something like flux jobs -i <hostlist> would make sense.

Design-wise, it arguably makes sense assume -A but that breaks the current paradigm.

[grondo]

Note there is sort of a solution to this proposed in #5711. It adds a query syntax to the -f, --filter option that would allow you to get this result with flux jobs -f host:<hostlist> -f running. That PR has never been merged because there isn't good consensus on the interface.

Perhaps filtering flux jobs by hostlist would be common enough that it should get its own option. I like your idea of using the -i, --include=hosts|ranks from the flux resource utilities, though I also wonder if that could be confusing in flux jobs which isn't listing resources (and not all jobs even have had resources assigned).

Anyway, we could support this option and and it with other filter options including any result of an advanced filter from -f, --filter= in the future.