make benchmarks supp. figure

[sg-s]

No description provided.

[sg-s]

i understand why the exponential euler doesn't converge exactly to the ode23t solution

that's because the ode23t solution isn't correct! when i change the tolerance of the ode solver, the solution changes a little bit. i'm not very confident that

1. our exponential euler method is stable AND accurate
2. we don't have the weird issues that certain other people have

[sg-s]

but this makes it tricky to make a figure that shows all of this