las3_pub / predictable_parallel_patterns

The current solution is not clean, but allows us to change error handling later on without missing any point in the programm where we would want to use it.

This means that high level tasks can be stolen and lays the groundwork for implementing different tasks like classic work stealing.

Right now no work can be spawned, we simply have a proove of concept that we start up each thread, work for the master to finish and then synchronize back to the main thread.