Expose time-series and access log metric to measure queueing latency at Envoy

*Title*: *Expose time-series and access log metric to measure queueing latency at Envoy*

*Description*:
To better attribute and breakdown e2e latency and correctly tune Envoy, we find it important to understand the queueing latency at Envoy for requests and responses. This will help us find out the proportion of time that a request or response spent locally at Envoy, and how that compares to the e2e latency.

[optional *Relevant Links*:]
>Any extra documentation required to understand the issue.