Add ports to ResumeProgram ports and add ResumeFailProgram

[sjpb]

Improves support for autoscaling:

• Modifies ResumeProgram: If a port is found on the specified network with the same name as the node, the instance is attached to that port. Allows for using predefined ports for e.g. non-default binding configurations or no DNS,
• Adds ResumeFailProgram: Handles cases for failed Resume (nodes -> DOWN where:
  • Instance not created: Node set to resume state to allow Slurm to retry that node (as required by scheduler)
  • Instance created but in error state due to "not enough hosts available": instance deleted, node set to resume state.

This also does a big refactor to remove duplicated code and allow use of suspend() and resume() functions from other scripts.