Add embarrasingly parallel example to doc

[vmlaker]

Add to documentation an example of how to distribute an embarrasingly parallel problem to mulitple cores. Consider a cascade of concentric for-loops, marshalling the outer for-loop to workers.