Follow-up review comments from 3066

Improve monitoring e2e test.

Up cmd/e2e timeout

Fix binary detection on GKE

* kubectl is kubectl.sh in the e2es, so not needed in the path
* Use the environment variable, and try to make sure the preview
component is installed.

Attempt to fix https://github.com/GoogleCloudPlatform/kubernetes/issues/3053

Do not use namespace in url paths pre v1beta3 from client

Fix the conversion routine for PodStatus to PodPhase 

Fix kube-controller-manager failing to start because "Jinja variable 'cloud_config' is undefined"

Adding skeleton swagger UI

Add a PodUnknown phase and make ListPods return even when there are errors

Update the Vagrant getting started guide

Replace kubecfg with kubectl in setup scripts

Remove debug print in kubectl.sh

Replaces two instances of kubecfg with kubectl in cluster startup and
validation scripts.

Removes a debug print that (it seems to me) was added in erro
as part of PR #3046 / commit c3efef86.

Fixes two instances that reference the deprecated kubecfg.sh tool.

Update the kubernetes_vagrant_auth example to more accurately reflect
what happened when I ran the scripts on my local machine.

Update the aws docs with a pointer to instance profile and roles.

Fix up AWS support...

obtaining info for some pods.

2. Updating datafile.go to include it.
3. Updating master,go to use it.

Allow parameterization by environment in cluster/gce/config-test.sh

Between this and careful use of "export HOME=/somewhere_else_for_auth/",
you can effectively get ghetto CLI profiles (for GCE tests).

bump(github.com/spf13/cobra):e1e66f7b4e667751cf530ddb6e72b79d6eeb0235

Add Host to PodSpec and add a predicate to make the scheduler work.

better logging

Specify AWS_DEFAULT_REGION to match configured region