Slurm node commands

Add one or more node names below — bracket expansion like gpu-[01,03-05] works, and you can paste a comma- or whitespace-separated list. The commands below regenerate as you add or remove nodes.

The Safety check script is the one to use when a cluster wakes up with dozens of drained nodes after an outage — it splits them into "drain reason was a transient pre-hook failure (safe to resume)" vs "real reason (don't touch yet)".