We use active health checking and observe an issue where endpoints that have been removed from service discovery are stuck in "pending dynamic removal state" and some requests are routed to them (they still pass HC). It would be useful to be able to customise the setting ignore_health_on_host_removal on cluster level: https://www.envoyproxy.io/docs/envoy/latest/api-v3/config/cluster/v3/cluster.proto.

[optional Relevant Links:]

Any extra documentation required to understand the issue.