The add RDS instance menu is populated dynamically by querying AWS using a service user token. It is populated only with instance end points and not cluster endpoints. Presumably this decision was made based on clusters where multiple hosts may be contacted via that IP and the results skewed; this overlooks the possibility of a fail over instance which is not used for anything else. In this situation you are needlessly utilizing twice as much storage to monitor both endpoints.
What I would like is to be able to select the cluster endpoint, because both instances will have the same data values but only the active one will have activity of interest to monitor and the incoming workload should be the same regardless of which one is active. There may be a brief skew in the graphs as buffer warming occurs but it should settle quickly.
More useful yet would be an option to allow just entering a name or IP for a cloud managed instance, this would be useful for other providers as well that functionality to search and find the instances has not been implemented for yet, eg: GCP Cloud SQL.