Add three benchmarks for the framework

[twilker]

For such a framework benchmarks are quite important. For that I added three benchmarks:

• Real World benchmark: editing 400 out of 2000 files
• Worst case scenario: Sequential execution (to measure the overhead based on total time executing one work unit)
• Base case scenario: 2000 randomized parallel executable workloads.

Optimize performance by removing unnecessary code.
There was a spinning thread which queued the messages. I am not sure why it was there. The only advantage of that is when it is important that messages are publish in the same order they are given to the message board. But that is no real scenario.

This resolves #21