Resource in job config will be changed after uploading to webportal

1. save this config to a yaml file. It claims 2 gpu, 10 cpu, and 6500 MB mem.

protocolVersion: 2
name: hello-world-job
type: job
jobRetryCount: 0
prerequisites:
  - type: dockerimage
    uri: 'openpai/standard:python_3.6-tensorflow_1.15.0-gpu'
    name: docker_image_0
taskRoles:
  taskrole:
    instances: 1
    completion:
      minFailedInstances: 1
      minSucceededInstances: -1
    taskRetryCount: 0
    dockerImage: docker_image_0
    resourcePerInstance:
      gpu: 2
      cpu: 10
      memoryMB: 6500
    commands:
      - 'git clone https://github.com/tensorflow/models'
      - cd models
      - git reset --hard 62bf6fc0bb2b41a4fb851909c132647a7e0774b0
      - cd research/slim
      - >-
        python download_and_convert_data.py --dataset_name=cifar10
        --dataset_dir=/tmp/data
      - >-
        python train_image_classifier.py --dataset_name=cifar10
        --dataset_dir=/tmp/data --max_number_of_steps=1000
defaults:
  virtualCluster: default
extras:
  com.microsoft.pai.runtimeplugin:
    - plugin: ssh
      parameters:
        jobssh: true

2. upload the yaml config to webportal, the resource has been changed to

Reproduce the issue by uploading the YAML configuration through the webportal and compare its resource values before and after upload. Trace where the webportal parses or transforms resourcePerInstance, then verify that the submitted gpu, cpu, and memoryMB values remain unchanged; no source file or test is named in the issue.