During the flush cycle batches are handed over to a the flush thread pool.
Batches are flushed when the size threshold is breached, or if the TTL expires. However, the TTL is only checked when a new Op is added.
During periods of inactivity we notice inaccurate counts which is undesirable.
It would be good if the TTL flush occurred even when no new Ops are incoming.

I would propose adding a monitor thread that checked every minute or so if there were any flushes pending. Thoughts?

[we need this for graceful shutdown reasons - currently we loose batches in flight]