[sled-diagnostics] chunk the number of pids for ptool commands

[papertigers]

This PR is on top of #8151 and is also ment to help address #8166.
It chunks up the number of PIDs passed to the various ptool commands such as pstack and pfiles

[papertigers]

I am inclined not to merge this because it means any transient process will mark a whole chunk of PIDs via pstack/pfiles/pargs as failed when only one or more PIDs are no longer around.

Open to thoughts here...but we know rust 1.87 helps improve things immensely.

[smklein]

I am inclined not to merge this because it means any transient process will mark a whole chunk of PIDs via pstack/pfiles/pargs as failed when only one or more PIDs are no longer around.

Open to thoughts here...but we know rust 1.87 helps improve things immensely.

I'm experimenting with this locally, with:

pfexec pstack <pid that exists> <pid that doesn't exist> <pid that exists>

I'm seeing valid results for all "real pids" go to stdout, and error messages go to "stderr". But a missing pid doesn't seem to inhibit the rest of the collection from occurring.

[papertigers]

I'm seeing valid results for all "real pids" go to stdout, and error messages go to "stderr". But a missing pid doesn't seem to inhibit the rest of the collection from occurring.

That's right but the command execution is looking at the exit status. So the data will be there but it will be considered failed.