cmd/pomerium : restore /ping without hostname (#37)

Re-adding the hostless /ping for the authenticate service. Without these 4 lines, Pomerium fails to stay running when using the helm chart that's checked into the repo; the Kubernetes liveness/readiness probes don't see the pod as successfully running, and Kubernetes goes into an endless restart loop.

I have tried the following:

Adding httpHeaders: {name: "Host", value: "insert_redirect_uri_hostname_here"} to both probes to try to force it to recognize. Pomerium fails to stay running.
Removing the probes from the deployment definition. Pomerium successfully stays running and responds correctly. Thus it's certain that it's one of the probes failing.
Modifying the code here to log the request.host golang reports. It reports "10.x.x.x:443". (Actual pod internal IP address for the 10.x.x.x). That won't match the authHost.

cmd/pomerium/main.go

@@ -68,6 +68,11 @@ func main() {
 
 	topMux := http.NewServeMux()
 	if authenticateService != nil {
+		// Need to handle ping without host lookup for LB
+		topMux.HandleFunc("/ping", func(rw http.ResponseWriter, _ *http.Request) {
+			rw.WriteHeader(http.StatusOK)
+			fmt.Fprintf(rw, "OK")
+		})
 		topMux.Handle(authHost+"/", authenticateService.Handler())
 	}
 	if proxyService != nil {