It is currently difficult to tell which clients are requesting entropy
from the daemon and how much they are requesting, along with whether the
request could be completely satisfied.

This would be useful for capacity planning, as at the moment the only
way for me to know if my entropy keys are overused is:

- wait until clients start exhausting their entropy
- check there's no problem local to the client
- start sniffing network and/or tracing ekeyd-egd-linux process to see
  who is requesting what.

Exposing some stats (even just logging) would make life easier.