Fix Helm chart service labels

[stevehipwell]

Description of changes:

This PR undoes the change made in #604 and fixes the actual issue I introduced in the chart refactor work.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[stevehipwell]

/assign @bwagner5

[snay2]

This is defined here, for reference:

aws-node-termination-handler/config/helm/aws-node-termination-handler/templates/_helpers.tpl

Lines 71 to 74 in 1c61c2d

	{{- define "aws-node-termination-handler.selectorLabelsDeployment" -}}
	{{ include "aws-node-termination-handler.selectorLabels" . }}
	app.kubernetes.io/component: deployment
	{{- end -}}

The important point is that it defines component to be deployment.

[snay2]

Just to clarify that my understanding of this change: the bug previously was that component was set to aws-node-termination-handler, which is not correct. The value should have been daemonset or deployment (depending on the way NTH runs in the cluster). Your solution is to remove this from the generic labels and instead assign it in the supersets of labels called labelsDeployment and labelsDaemonset, and then to use the unrelated selectorLabelsDeployment for the service monitor template, which separately sets component to deployment.

Do I understand correctly?

[stevehipwell]

Yes that is correct. When I refactored the chart I accidentally left the component in the main labels function and forgot to add the component labels functions. Without these functions the component was incorrectly set in the resource labels but the only bug would be the ServiceMonitor as that's the only selector working on these labels.