Processor Self-Scheduling for Multiple-Nested Parallel Loops