Description:

We've seen some issues on server side where a single session is pinning too much cache causing eviction stuck. This ticket is to investigate anything should be returned if a session is holding too much cache, and how to define that, so we can warn the server when necessary.

A suggestion could be utilising the existing hazrd pointer max array length, when a session's hazard pointer is considered as too many under the current WT cache, we should log the stat or warn the server.

Definition of done:

Add clean bytes, dirty bytes, update bytes for session.