Enable saving the entire simulation state in memory and restoring it again

[kyllingstad]

I had originally intended this as a subtask of #757, which is about persisting the entire simulation state to disk, but after working on #756, I've found that it makes more sense to have it as a separate issue.

In essence, this issue is about implementing similar functionality as in #756, only on the full-system level. Here, we also need to export the state of the algorithm.