It should be straightforward to compute the global minimum time step for a simulation, assuming that we know the max thermal conductivity, the minimum coarse mesh element size, and the maximum number of refinements. This would allow us to make it so the user doesn't need to specifically pick the time step size, which would make things easier. We could add a user-specified safety factor so they can decrease the time step if desired.

Later on we could do a more sophisticated version of this where we adapt based on the layer that we're on, but that's out of scope for now.