Similar to ACM-18001, several alerts have issues where we're taking a rate of a given metric and expect it to come above 10. Usually these metrics only increase rarely (for example once every 5 minute/spoke), which means the alert never fire unless the amount of spokes are super high.

A review of these should be done, and see if we want to fix them, or drop them altogether. Ideally the alerts would fire on error percentages (i.e if 20% of all requests fail) instead of using a static rate.

Alerts affected:

• ACMMetricsCollectorFederationError
• ACMMetricsCollectorForwardRemoteWriteError
• ACMUWLMetricsCollectorFederationError
• ACMUWLMetricsCollectorForwardRemoteWriteError
• https://github.com/stolostron/multicluster-observability-operator/blob/2f4693b554510a12daa162da6018bcc7793568b4/operators/endpointmetrics/pkg/collector/metrics_collector.go#L447 (metrics collector ones on spokes)