It turns out I had a deprecated check left over when determining a starting point in the test_get_integer_time_begin function.
I first pick some random point on the timeline which is a viable time_begin, and then determine a random point on the timeline at a later integer-time and try to reconstruct the originally picked time_begin. While picking the original time_begin, I threw an error if time_begin == max_nr_steps, which was an oversight. (Particularly so since I later explicitly also tests what happens if I provide the function with a current time that is > max_nr_timesteps.)