helm: Do not parse initial 'helm get' section as a manifest

[tstromberg]

This removes the scary error decoding parsed yaml: Object 'Kind' is missing in message that appears with every Helm deployment.

This log message is caused due to skaffold parsing the front-matter of helm get output as if it was a Kubernetes manifest. This PR skips the initial section for object parsing, and raises the warning level for subsequent sections from info to error.

Fixes #3642

Using examples/helm-deployment with skaffold run -v info, you can see the difference:

At head

INFO[0030] Building helm dependencies...                
NAME: skaffold-helm
LAST DEPLOYED: Thu May 21 11:44:36 2020
NAMESPACE: default
STATUS: deployed
REVISION: 1
TEST SUITE: None
INFO[0031] error decoding parsed yaml: Object 'Kind' is missing in 'NAME: skaffold-helm
LAST DEPLOYED: Thu May 21 11:44:36 2020
NAMESPACE: default
STATUS: deployed
REVISION: 1
TEST SUITE: None
USER-SUPPLIED VALUES:
image: skaffold-helm:9eeb2cd8a4cb941fc16f4fc1796d713da20308e5f391cd10a35b1aa4bb8f6e2a

COMPUTED VALUES:
image: skaffold-helm:9eeb2cd8a4cb941fc16f4fc1796d713da20308e5f391cd10a35b1aa4bb8f6e2a
ingress:
  annotations: null
  enabled: true
  hosts: null
  tls: null
replicaCount: 1
resources: {}
service:
  externalPort: 80
  internalPort: 80
  name: nginx
  type: NodePort

HOOKS:
MANIFEST:
' 
INFO[0031] Deploy complete in 1.403286918s              
Waiting for deployments to stabilize...

With this PR

INFO[0000] Building helm dependencies...                
NAME: skaffold-helm
LAST DEPLOYED: Thu May 21 11:41:40 2020
NAMESPACE: default
STATUS: deployed
REVISION: 1
TEST SUITE: None
INFO[0001] Deploy complete in 1.149852761s              
Waiting for deployments to stabilize...

[briandealwis]

Is there any value in logging that section header?

I'm (very mildly) worried that Helm 4 or 5 might starting emitting valid k8s manifests and we'll throw this first element away. Would it make sense to instead try parsing it and instead ignore the error if it's the first object?

[briandealwis]

@tstromberg were you still interested in getting this in?

[nkubala]

@tstromberg this is starting to get stale, could you have a look at the failing unit tests?

[gsquared94]

@tstromberg this is starting to get stale, could you have a look at the failing unit tests?

@tstromberg ping

[briandealwis]

I have a simpler implementation.