More timings

[rykerfish]

I've added timings for the rheology and diffusion sections that were used to expand the benchmark section of the paper.

This also includes a change to the original benchmarks where particles are now placed randomly instead of in an FCC lattice. The problem with the lattice was it greatly undersold DPStokes since the lattice required large domains where DPStokes begins to slow down. Instead, random packings create denser suspensions where DPStokes obviously outperforms the other algorithms when we are into the range of tens of millions of particles.