Since flux depends on core area, not gap, it doesn't matter!
Indeed, if you set up a saturation tester (typically by driving a pulse into the winding -- just as the simulation above), and you use an inductor on a gapped core, then you can adjust the gap manually, in real time, while testing. As you adjust the gap length, the slope of the current varies (because of the inductance -- more gap = less inductance = more magnetizing current), but the time where saturation begins will remain nearly constant. Because, for constant applied voltage, the same flux is reached at that moment, (almost) independent of gap (or inductance or current draw).
Conversely, because the resistance of a real circuit is nonzero, a gapped core is less likely to saturate in a full-wave circuit. Why? For the same reason the current eventually reaches zero on the diode-clamped inductor, or in a short-circuited (non superconducting) inductor. The increased magnetizing current causes more voltage drop across circuit resistance, which drops a voltage, and thus a flux, that's proportional to current. It's a negative feedback process.
When you solve the equation for that feedback process -- that is, where applied voltage is dependent on flux -- you get an exponential decay type solution: the inductor-resistor time constant.
Test this in the simulator by connecting a voltage source to an inductor, and setting the inductor for a nonzero initial current, and a nonzero series resistance. You will see the average drifts back to zero, over time!
The rate at which current decays is the L/R time constant. Written as a function, I(t) = I(0) exp(-t * R/L).
Tim