Currently metrics and alerts are only enabled in the openshift bundle, because for OpenShift the `ServiceMonitor` and `PrometheusRule` CRDs are always installed. If CRs of these types are part of the bundle but the CRD doesn't exist in the cluster, OLM will fail the entire installation.

There is a OLM enhancement proposal for optional components: https://github.com/operator-framework/enhancements/blob/master/enhancements/optional-manifest.md

Alternatively, we could install the `ServiceMonitor` and `PrometheusRule` objects after operator startup, depending on a (new) operator configuration option.

https://github.com/grafana/tempo-operator/issues/491