Multithreaded APR?

I was wondering if it might be possible for the `update_apr` function to parallelize the split and merge loops using OpenMP. 
These steps can become a noticeable bottleneck, especially when splitting a large number of particles that constantly enter/exit the refinement regions,