Rename instance label to aws_resource

[fheinecke]

The instance label typically stores an ARN or resource ID for a resource being monitored. This poses a problem when using the prometheus exporter with prometheus, because Prometheus automatically adds the instance label to metrics in a way that is particularly difficult to work around.

This PR renames the instance label to aws_resource, which should more accurately describe the label's purpose, and not collide with Prometheus' instance label.

[taraspos]

Hmm... I think prometheus might automatically rename existing instance label to exported_instance and it's already being used in some of the dashboards:

• https://github.com/search?q=repo%3Agravitational%2Faws-quota-checker%20exported_instance&type=code

• https://github.com/prometheus/prometheus/blob/6683895620ac5d6e20b58ccbff8b20604e55881a/docs/configuration/configuration.md?plain=1#L201-L204

[fheinecke]

I think prometheus might automatically rename existing instance label to exported_instance and it's already being used in some of the dashboards

It depends on how this is deployed, and whether honor_labels is true or false (as you linked). IMO it's a better approach to not collide with a well-known Prometheus label, even if there is a workaround.

The Helm chart PR fixes this in the dashboard (tested and confirmed working), but I'd be glad to pull that change into this PR if you like, if this approach works for you.

[taraspos]

We need to remember to update some of our custom dashboards when upgrading aws-quota-checker to latest version