Expose Linode Client Prometheus metrics (#302)

* Expose Linode Client Prometheus metrics

Add `--authorization-always-allow-paths="/metrics"` to command line to
allow scraping these metrics.

In addition to Linode Client metrics, standard controller and worqueue
metrics are visible by default.

To disambiguate CCM node controller work queue name, it's explicitly
named "ccm_node". That is, "node", "service" are upstream work
queues of k8s cloud-provider service controller and "ccm_node" is
the work queue of Linode CCM node controller.

Fixes #296

* Ignore coverage for client_with_metrics.go

* Add allowUnauthorizedMetrics Helm value

* Move metrics docs to getting-started sections

* Instrument healthChecker with Prometheus metrics

* Update newHealthChecker tests

* Preserve initialization order

* Remove unused timeout variable in healthChecker

* Remove unnecessary err return in newHealthChecker

Author: rsienko

cloud/linode/client/client.go

@@ -1,6 +1,7 @@
 package client
 
 //go:generate go run github.com/golang/mock/mockgen -destination mocks/mock_client.go -package mocks github.com/linode/linode-cloud-controller-manager/cloud/linode/client Client
+//go:generate go run github.com/hexdigest/gowrap/cmd/gowrap gen -g -p github.com/linode/linode-cloud-controller-manager/cloud/linode/client -i Client -t ../../../hack/templates/prometheus.go.gotpl -o client_with_metrics.go -l ""
 
 import (
 	"context"
@@ -10,6 +11,7 @@ import (
 	"os"
 	"time"
 
+	_ "github.com/hexdigest/gowrap"
 	"github.com/linode/linodego"
 	"k8s.io/klog/v2"
 )