Add Logstash Telemetry

[robbavey]

This PR adds telemetry related to Logstash

[robbavey]

PR ready for review

[kaisecheng]

remove it?

[robbavey]

This one was caught by the linter, and is related to loop variable semantics in go - there is a great summary of the issue (and proposal for "fixing" it in future versions of golang) here. I don't actually think that that behaviour will bite us here, given what we are doing, but I'm still happy keeping this "odd" looking assignment in here and keep the linter satisfied

[robbavey]

@pebrc @barkbay This is ready for review by the ECK team

[thbkrkr]

LGTM, left a few nits.

It's not from your PR, but I realize that we started having duplicates and making the stats slightly different from app to app, causing us to miss some stats (stack mon for Kibana, or helm for agent and beats).

I think we could improve that by creating some reusable structs like this.

type baseStats struct {
	ResourceCount int32 `json:"resource_count"`
	PodCount      int32 `json:"pod_count"`
}

type helmStats struct {
	HelmManagedResourceCount int32 `json:"helm_resource_count"`
}
type stackMonStats struct {
	StackMonitoringLogsCount    int32 `json:"stack_monitoring_logs_count"`
	StackMonitoringMetricsCount int32 `json:"stack_monitoring_metrics_count"`
}

func logstashStats(k8sClient k8s.Client, managedNamespaces []string) (string, interface{}, error) {
	logstashStats := struct {
		baseStats
		helmStats
		stackMonStats
		ServiceCount     int32 `json:"service_count"`
		PipelineCount    int32 `json:"pipeline_count"`
		PipelineRefCount int32 `json:"pipeline_ref_count"`
	}{}
...

I made a sketch here.

It's not completely trivial so it can be done in another PR, unless you're motivated to do it in this one :).

[robbavey]

@thbkrkr Ready for another round