Investigate if this is a cause for limited transaction throughput, especially we often see smaller blocks, after larger blocks.
In the current implementation we are waiting until the last block was imported before we send a further contribution.
We need to check if this is a pattern that transactions are contributed after they already become included.
maybe transaction queue not up to date yet ?
this could explain "echo" blocks.