Handle RESOURCE_BUSY responses from CloudControl API

[tintoy]

Sometimes CloudControl will indicate a transient problem due to resource contention by returning RESOURCE_BUSY. If that happens we should be able to retry operations.

For example, deploying a VLAN while another VLAN is being deployed for the same organisation will result in RESOURCE_BUSY. The provider should be able to retry the operation until it succeeds or times out.

• [x] Retry mechanism
• [x] Provider settings
  • [x] Retry period
  • [x] Retry timeout
• [x] ddcloud_vlan
  • [x] Create
  • [x] Delete
• [x] ddcloud_networkdomain
  • [x] Create
  • [x] Delete
• [x] ddcloud_nat
  • [x] Create
  • [x] Delete
• [x] ddcloud_firewall_rule
  • [x] Create
  • [x] Delete

Let's do something like:

var apiClient *compute.Client

var result string
err := retry.Action("Deploy VLAN", 3 * time.Minute, func(context retry.Context) {
  var e error
  result, e := apiClient.DeployVLAN(name, description, baseAddress, prefixSize)
  if e != nil {
    context.Fail(e)

    return
  }

  if isBusy(result) {
    context.Retry()
  }
}

if err != nil {
  return "", err
}

return result, nil

[tintoy]

Have given this a quick test-drive, @mingsheng36, and it looks promising :)