Connection refused message when deploying to stopped minikube cluster is confusing

From friction log:
https://docs.google.com/document/d/1h2VKg3WZ9sHgNWXvmliq0UVJGUcdrbECSeYrjZYuVVw/edit

Minikube was in a stopped (or possibly paused) state. And Skaffold spit out:

```
/Users/russellwolf/Library/Application Support/cloud-code/bin/versions/2124dec347bde76999f47dad22d7f4c293b3cdf367dfbf3b6106a6b4a730c17d/skaffold dev --filename skaffold.yaml --label ide=Idea --label ideVersion=2020.1.3.0.0 --label ijPluginVersion=unknown --default-repo gcr.io/next-london-cloudcode --rpc-port 50051 --port-forward=true --enable-rpc=true
Listing files to watch...
- java-guestbook-backend
- java-guestbook-frontend
Generating tags...
- java-guestbook-backend -> gcr.io/next-london-cloudcode/java-guestbook-backend:latest
- java-guestbook-frontend -> gcr.io/next-london-cloudcode/java-guestbook-frontend:latest
Checking cache...
- java-guestbook-backend: Found Remotely
- java-guestbook-frontend: Found Remotely
Tags used in deployment:
- java-guestbook-backend -> gcr.io/next-london-cloudcode/java-guestbook-backend:latest@sha256:605dc171dcc7b2f8b9e18dfb4eaed2980c94425b360755368e7868667838ac17
- java-guestbook-frontend -> gcr.io/next-london-cloudcode/java-guestbook-frontend:latest@sha256:ca2e1e15f74ef0a65c062cffeddc6b33794e6d3c79ce41b6f649a196de96db59
exiting dev mode because first deploy failed: unable to connect to Kubernetes: Get "https://127.0.0.1:32772/version?timeout=32s": dial tcp 127.0.0.1:32772: connect: connection refused
```

Instead `connect: connection refused`, Skaffold could maybe provide a nicer message (since it should know if it is a local minikube cluster or not). Something very roughly like: `minikube cluster is not running. Start minikube..`