feat: support custom branch for upscale

[jacderida]

• 2133a2d chore: remove remaining traces of network keys

  These were removed from antnode and are no longer required.

• ca32ee4 chore: use different version retrieving mechanism

  The versions for both antnode and antctl can just come from a random VM rather than the node
  registry.

• ca780ed feat: support custom branch for upscale

  The upscale does not build the binaries. Instead it uses the build from the original deployment. The
  binary option can be set to the same owner/branch reference as the previous deployment, and this
  means Ansible will pick up the URL where the custom binary was uploaded to.

• 07ef86b chore: remove redundant args from upscale cmd

  The --desired-auditor-vm-count and --downloaders-count arguments do not apply any more.

• 5ae4436 refactor: use shared mechanism for apply and plan args

  The terraform plan and terraform apply arguments should be identical to each other, such that
  the plan will be producing the same change set as the apply. There were different code paths for
  each of them, which meant they could diverge easily.

  Extract a full function for obtaining the resource value from the Terraform state. The closure was
  large, so it was making the enclosing function more difficult to read. The vm_size variable was
  renamed because it was misleading: the function retrieves the value of any resource, not just the
  size of a VM.

  BREAKING CHANGE: I removed the plan command as part of this refactor. This command on its own just
  didn't make a whole lot of sense. It makes more sense to be used with an upscale.

• 0d057a4 feat: obtain machine sizes from previous deploy

  It's possible to launch the original deployment with values different from those in the tfvars file.
  If you then try to upscale, Terraform will then want to apply the values from the tvfars file, which
  could be smaller or larger.

  The sizes are therefore obtained from the original deployment by inspecting the current state.