Minor updates to K8s deployment and Helm chart

[alexei-led]

This PR includes few small changes to basic K8s deployment and Helm chart:

1. Use kube-system namespace for CRD workflow-controller and ServiceAccount
2. In order to fetch latest image, the imagePullPolicy should be set to Always
3. Current workflow-controller does not support running more than 1 replica, due to design (all replicas will get K8s events and will try to create jobs).
4. When using K8s GKE cluster, enabling RBAC requires defining an additional CRB to access K8s dashboard. I've added gke-kube-dashboard-rbac.yml for this

p.s.: workflow-controller works nicely on a multi-node K8s GKE cluster with hundreds of running "stress" steps.

[codecov-io]

Codecov Report

Merging #23 into master will not change coverage. The diff coverage is n/a.

@@          Coverage Diff           @@
##           master     #23   +/-   ##
======================================
  Coverage    23.4%   23.4%           
======================================
  Files           8       8           
  Lines         833     833           
======================================
  Hits          195     195           
  Misses        623     623           
  Partials       15      15

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update f0f1b6b...4ffc103. Read the comment docs.

[sdminonne]

@alexei-led I think this one can be merged but I'm double checking with @clamoriniere1A

[alexei-led]

@sdminonne thank you.

p.s.: maybe worth to open Gitter room for this project, to simplify IM communication?

[sdminonne]

https://gitter.im/workflow-controller/Lobby