Add a case study on kernel performance for PIC code field solve like kernel....

Add a case study on kernel performance for PIC code field solve like kernel. This case study shows the affects of loop unrolling on run time performance on the kernels. Partial loop unrolling on the GPU shows a significant performance imporvement on GPU. This may be an effective way to impliment the field solve on the GPU without needing to create a loop or sum nodes in the graph. However, we may need to impliment MAX, MIN nodes or If/Else nodes to deal with functionality involving grid bounds and particle reinjection.