Bump dashboard to 1.8.3 to work with Kubernetes 1.9.4

[Cynerva]

Dashboard broke when we released 1.9.4. Pod gets stuck in a crash loop with this log output:

$ kubectl logs kubernetes-dashboard-7b7b5cd79b-95zqt -n kube-system
2018/03/12 22:15:43 Starting overwatch
2018/03/12 22:15:43 Using in-cluster config to connect to apiserver
2018/03/12 22:15:43 Using service account token for csrf signing
2018/03/12 22:15:43 No request provided. Skipping authorization
2018/03/12 22:15:43 Successful initial request to the apiserver, version: v1.9.4
2018/03/12 22:15:43 Generating JWE encryption key
2018/03/12 22:15:43 New synchronizer has been registered: kubernetes-dashboard-key-holder-kube-system. Starting
2018/03/12 22:15:43 Starting secret synchronizer for kubernetes-dashboard-key-holder in namespace kube-system
2018/03/12 22:15:44 Initializing JWE encryption key from synchronized object
2018/03/12 22:15:44 Creating in-cluster Heapster client
2018/03/12 22:15:44 Auto-generating certificates
2018/03/12 22:15:44 [ECDSAManager] Failed to open dashboard.crt for writing: open /certs/dashboard.crt: read-only file system

This was fixed in kubernetes-dashboard 1.8.3, here: kubernetes/dashboard#2795

Lightly tested, seems to work.

[Cynerva]

Self-merging to expedite this. @wwwtyro review please.

[wwwtyro]

lgtm