Parallelizing DSW Trees

A parallelization of the Day-Stout-Warren algorithm for balancing binary trees. As its input, this algorithm takes an arbitrary binary tree and returns an equivalent tree which is balanced so as to preserve the θ(log(n)) lookup time for elements of the tree. The sequential Day-Stout-Warren algorithm has a linear runtime and uses constant space. This new parallelization of the Day-Stout-Warren algorithm attempts to do the same while providing a speedup which is as near as possible to linear to the number of processing elements. Also, ideally it should do so in an online fashion, without blocking new reads, inserts, and deletes.