oc adm upgrade recommend should retrieve alerts from the cluster (similar to how oc adm upgrade status already does), and inject them as conditional update risks (similar to how OTA-902 / cvo#1907 injected Upgradeable issues). The set of alerts to include is:

• All critical alerts, because that's explicitly selected in OCPSTRAT-1834.  This includes ClusterOperatorDown, which overlaps with the existing ClusterVersion Failing condition, but 🤷.  It also includes the PodDisruptionBudgetLimit alert which is part of PDB coverage.
• PDB coverage:
  • warning PodDisruptionBudgetAtLimit.
• Image registry access:
  • warning KubeContainerWaiting.
• Nodes:
  • warning KubeNodeNotReady, KubeNodeReadinessFlapping, and KubeNodeUnreachable.

Definition of done / test-plan:

1. Find a cluster with both update recommendations and some of the mentioned alerts firing.
2. Run OC_ENABLE_CMD_UPGRADE_RECOMMEND=true OC_ENABLE_CMD_UPGRADE_RECOMMEND_PRECHECK=true oc adm upgrade recommend.
3. Confirm that the command calls out the expected alerts for all update targets.