Feature: resync nodes with CycleCloud after failures

When a failed suspend, or to a lesser extent, resume, happens we need a tool to quickly resync the state in CycleCloud with that of Slurm.

This most likely needs to be interactive by default, with only automated actions taking place if something like --force is passed.

Start by locating the existing suspend and resume failure-handling entry points and how Slurm and CycleCloud state are queried. Define the interactive workflow and --force behavior, then verify that a failed suspend or resume can be reconciled without automated actions unless --force is provided.

Written by the indexing model from the issue text.