bounded_runs_iter simulates a strange number of time steps

[Ratfink]

The bounded_runs_iter function takes parameters n and t, specifying the number of time steps per call to bounded_runs and the number of iterations. But currently it actually simulates t+1 iterations, which is just a strange decision on my part. We should fix it so the number of time steps is just n*t instead of n*(t+1).

[Ratfink]

In fact, it would be better to specify n, the number of steps per iteration, and H, the overall time horizon. That way, a user could more easily adjust the parameter n while keeping the number of steps the same. t feels more like an implementation detail anyway.