model resolution seems rather unused

[clonker]

The virtual resolution method on a cloe Model does not really influence the simulation step size but rather just returns it (or does it?). In any case the docstring indicates IMO that it would effect the simulation loop integration step size. I think the docstring should be updated or perhaps the method even removed/replaced with a non-virtual counterpart that returns the configured value.

[cassava]

This is true: we don't actually have the logic in place to respect this.

But we wanted to put it in the API because we anticipated it. We have been waiting for someone to need it.