core: fix pthreads performance

OpenCV pthreads-based implementation changes:
- rework worker threads pool, allow to execute job by the main thread too
- rework synchronization scheme (wait for job completion, threads 'pong' answer is not required)
- allow "active wait" (spin) by worker threads and by the main thread
- use _mm_pause() during active wait (support for Hyper-Threading technology)
- use sched_yield() to avoid preemption of still working other workers
- don't use getTickCount()
- optional builtin thread pool profiler (disabled by compilation flag)

modules/core/src/parallel_impl.hpp

@@ -0,0 +1,17 @@
+// This file is part of OpenCV project.
+// It is subject to the license terms in the LICENSE file found in the top-level directory
+// of this distribution and at http://opencv.org/license.html.
+#ifndef OPENCV_CORE_PARALLEL_IMPL_HPP
+#define OPENCV_CORE_PARALLEL_IMPL_HPP
+
+namespace cv {
+
+unsigned defaultNumberOfThreads();
+
+void parallel_for_pthreads(const Range& range, const ParallelLoopBody& body, double nstripes);
+size_t parallel_pthreads_get_threads_num();
+void parallel_pthreads_set_threads_num(int num);
+
+}
+
+#endif // OPENCV_CORE_PARALLEL_IMPL_HPP