Rework lock-free-work-stealing-deque to be simpler and fit 'classic' approach.

app/invoke_parallel/main.cpp

@@ -6,7 +6,6 @@
 #include <vector>
 
 static constexpr int CUTOFF = 16;
-static constexpr int NUM_ITERATIONS = 1000;
 static constexpr int INPUT_SIZE = 2064;
 typedef std::vector<std::complex<double>> complex_vector;
 

lib/pls/src/internal/scheduling/fork_join_task.cpp

@@ -23,7 +23,6 @@ void fork_join_sub_task::execute() {
   PROFILE_WORK_BLOCK("execute sub_task")
   tbb_task_->currently_executing_ = this;
   if (executed) {
-    int my_id = base::this_thread::state<thread_state>()->id_;
     PLS_ERROR("Double Execution!")
   }
   executed = true;
@@ -64,9 +63,7 @@ fork_join_sub_task *fork_join_task::get_local_sub_task() {
 }
 
 fork_join_sub_task *fork_join_task::get_stolen_sub_task() {
-  auto tmp = deque_.save_state();
-  auto result = deque_.pop_head();
-  return result;
+  return deque_.pop_head();
 }
 
 bool fork_join_task::internal_stealing(abstract_task *other_task) {