slurm: prevent nodehost column truncation in get_cluster_info

[samoz83]

Please review our Contributing Guide before submitting a pull request.

What does this PR do?

Fixes OodCore::Job::Adapters::Slurm::Batch#get_cluster_info so that GPU
counts are parsed correctly on clusters with node FQDNs longer than
20 characters.

sinfo -O uses a default width of 20 characters for the nodehost field.
When a hostname exceeds that width, sinfo truncates it and emits no
separating whitespace before the next column, causing line.split to
return 3 tokens instead of 4. The gres and gresused columns are then
read from the wrong indices: total_gpus ends up summing GresUsed (the
in-use count) and active_gpus tries to parse the state string
("mixed", "idle", …), which the GPU regex never matches, so it
evaluates to 0. The dashboard's System Status widget consequently renders
"GPUs Available" as the number of GPUs currently in use.

The fix sizes the nodehost column dynamically from sinfo -ho %n,
mirroring how gres_length is already calculated, so the format string
always reserves enough space to keep the columns separated regardless of
hostname length.

Related issue

Closes #945

Testing

• Tests included
• No tests needed — reason: ___

Checklist

• Follows project code style and conventions
• Documentation provided (if new feature, adapter or behavior change)
• This is a large feature and was discussed in an issue first (if applicable)

Anything else?

The fix has been validated on a real cluster running OOD 4.2.2 with
ood_core 0.31.1: applied as a dashboard initializer monkey-patch first
(to confirm), then upstreamed here. After applying, the System Status
widget shows the correct configured − in_use GPU count rather than the
in-use count.

[johrstrom]

I'm really unsatisfied with the fact that we make multiple calls to find the max length. Can't we just set the limit to some super high number like 100? This would save us the additional sinfo call. And yes we likely need to do it for the gres_length as well, just set it to some arbitrarily high number up front without having to make that sinfo call too.

[samoz83]

Pushed an update that drops both dynamic-sizing sinfo calls and uses fixed widths instead. Picked 512 for gres/gresused as it was double the longest gres string in the existing test fixture.

Also added a regression test so future refactors don't silently re-trigger the truncation case.

[johrstrom]

Thanks! This looks good to me.