The previous code to terminate the simulation was quite complex and
misleading.

The current one does essentially the same but stores another state to
make things simpler. An is_simulation_finished() function has been added
for this purpose. The main loop is now simply to wait until the
scheduler awknowledges the end of the simulation.