VNFD is presented in an arbitrary on horizon

[JangwonLee]

When a user creates a VNFD with a template of type dic, VNFD is presented on Horizon in an arbitrary order, non-identical to the order of the input file.

The result like below(by Horizon)
template: tosca-vnfd-alarm-scale.yaml

description: Demo example
metadata: {template_name: sample-tosca-vnfd}
topology_template:
  node_templates:
    CP1:
      properties: {anti_spoofing_protection: false, management: true}
      requirements:
      - virtualLink: {node: VL1}
      - virtualBinding: {node: VDU1}
      type: tosca.nodes.nfv.CP.Tacker
    VDU1:
      capabilities:
        nfv_compute:
          properties: {disk_size: 1 GB, mem_size: 512 MB, num_cpus: 2}
      properties:
        availability_zone: nova
        image: cirros-0.4.0-x86_64-disk
        metadata: {metering.server_group: SG1}
        mgmt_driver: noop
      type: tosca.nodes.nfv.VDU.Tacker
    VL1:
      properties: {network_name: net_mgmt, vendor: Tacker}
      type: tosca.nodes.nfv.VL
  policies:
  - SP1:
      properties: {cooldown: 120, default_instances: 1, increment: 1, max_instances: 3,
        min_instances: 1}
      targets: [VDU1]
      type: tosca.policies.tacker.Scaling
  - vdu_cpu_usage_monitoring_policy:
      triggers:
        vdu_hcpu_usage_scaling_out:
          action: [SP1]
          condition: {aggregation_method: mean, comparison_operator: gt, constraint: utilization
              greater_than 80%, evaluations: 1, granularity: 60, resource_type: instance,
            threshold: 80}
          event_type: {implementation: ceilometer, type: tosca.events.resource.utilization}
          metadata: SG1
          metric: cpu_util
        vdu_lcpu_usage_scaling_in:
          action: [SP1]
          condition: {aggregation_method: mean, comparison_operator: lt, constraint: utilization
              less_than 10%, evaluations: 1, granularity: 60, resource_type: instance,
            threshold: 10}
          event_type: {implementation: ceilometer, type: tosca.events.resource.utilization}
          metadata: SG1
          metric: cpu_util
      type: tosca.policies.tacker.Alarming
tosca_definitions_version: tosca_simple_profile_for_nfv_1_0_0

It caused by the property of dict. At first, after 'yaml.load()' the order has changed.
And 'yaml.safe_dump()' dumps changed dict order by alphabet. So the above result has created.

We need to make ordered_load and ordered_dump function to fix the VNFD order.