Add support to parse response headers from the task output

[PrasadG193]

Fixes https://github.com/fission/fission-workflows/issues/228 Example Usage:

• Pass complete list of output headers as input headers to a task

  inputs:
    headers: "{$.Tasks.taskid.OutputHeaders}" // Or "{outputHeaders('taskid')}"

• Pass only 'Foo' header from the output headers of a task as a input header

  inputs:
    headers:
      Foo: "{$.Tasks.taskid.OutputHeaders.Foo}" // Or "{outputHeaders('taskid').Foo}"

[PrasadG193]

@erwinvaneyk I noticed headers were missing from Workflows response also. I have added changes for the same.

[PrasadG193]

Hey @erwinvaneyk, sorry for the delay, I tried running test/e2e/buildtest.sh on my local setup with docker repo, it worked fine. Not sure what's going wrong on CI builds. Is it using right workflows docker images?

[erwinvaneyk]

Looking at the test logs, there indeed seems to be something going wrong in CI setup. I will investigate asap

[PrasadG193]

@erwinvaneyk Looking at the workflows deployment specs in helm charts,

# Source: fission-workflows/templates/deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: workflows
  namespace: fission
  labels:
    chart: "fission-workflows-0.6.0"
spec:
  replicas: 1
  selector:
    matchLabels:
      app: workflows
  template:
    metadata:
      annotations:
        prometheus.io/path: /metrics
        prometheus.io/port: "8080"
        prometheus.io/scrape: "true"
      labels:
        svc: workflows
        app: workflows
    spec:
      containers:
      - name: workflows
        image: "gcr.io/fission-ci/fission-workflows-bundle:ci-test"
        command: ["/fission-workflows-bundle"]
        args: [
          "--fission",
          "--internal",
          "--controller",
          "--api-http",
          "--api-workflow-invocation",
          "--api-workflow",
          "--api-admin",
          "--metrics",
        ]
        env: # TODO add dedicated NATS cluster (instead of reusing the mqtrigger)
        - name: FNENV_FISSION_CONTROLLER
          value: "http://controller.fission"
        - name: FNENV_FISSION_EXECUTOR
          value: "http://executor.fission"
      - name: jaeger-agent
        image: jaegertracing/jaeger-agent
        ports:
        - containerPort: 5775
          protocol: UDP
        - containerPort: 6831
          protocol: UDP
        - containerPort: 6832
          protocol: UDP
        - containerPort: 5778
          protocol: TCP
        args: [
          "--collector.host-port=jaeger-collector:14267"
        ]

imagePullPolicy: is not set. That means it is using IfNotPresent by default. So might be using older docker image for the deployment.

[erwinvaneyk]

@PrasadG193 - that could be it indeed 👍 I opened a quick PR for it since it might affect other PRs as well. (#237)

[erwinvaneyk]

Passed in https://travis-ci.org/fission/fission-workflows/jobs/488444845

Thanks for the great work Prasad! 🎉