There are existing filtering with metrics: https://docs.konghq.com/mesh/latest/policies/traffic-metrics/#filter-envoy-metrics. We want ways to further reduce the size of metrics. This is essential so we can avoid pulling large payloads from the /metrics endpoint from the data dog side. We get about 84k lines from one data plane and this continues to grow as more data planes are added and make it difficult to parse through so much logs.