Title: Collect gRPC statistics useful for autoscaling (active-requests, requests-time...)

Description:

Envoy is currently recording statistical data relating to gRPC:

https://www.envoyproxy.io/docs/envoy/v1.12.0/configuration/http/http_filters/grpc_stats_filter.html

<grpc service>.<grpc method>.success
<grpc service>.<grpc method>.failure
<grpc service>.<grpc method>.total
<grpc service>.<grpc method>.request_message_count
<grpc service>.<grpc method>.response_message_coun

But, thinking in autoscaling (like Kubernetes HPA), I miss information about
response times, number of active requests, etc...

At least, something like this:

<grpc service>.<grpc method>.activeRequests
<grpc service>.<grpc method>.upstream_rq_time  (histogram)