And anti-tank aircraft didn't come in at a low angle either. If you do, don't cry when you die.
That was because they were trying to hit the softer top armor, not in order to avoid the main gun.
Lets assume an IL2 flying at 200 mph and a tank is aiming its main gun at it. lets take IL2's length as the typical dimension for the target which is ~13m and the shooting distance as ~1000m. In terms of angular size, this is a fairly big target for a tank gun - at least a modern one, I don't know about WWII.
Lets check the time scales involved. If the IL2 is flying at 200 mph = 300 km/s (slow even for IL2) it crosses its own dimension in ~0.15 sec. Typical human reaction time (according to
http://www.humanbenchmark.com/tests/reactiontime/stats.php) is 0.215 sec with a standard deviation of about 0.035 (reading from the plot). Therefore, roughly 10% of the population may have good enough reflexes to consistently time the shot right - given they have a
perfect aim in space and a timing indication of when to pull the trigger.
Now, lets check the timing accuracy - There is no timing aid (lead indicator), at least not in a tank. The shell is subsonic (or is it?), so we'll assume 300 m/s ~Mach0.9. At 1000m this means ~3 seconds flight time to hit a target that will stay in the kill zone for 0.15 sec. You need a relative timing estimation of 0.05 accuracy. I don't have any data for that, but it would be interesting how accurate can a person estimate 3 seconds.
Anyway, what I get from this is that it is "humanly possible" to make the shot and probably the limiting factor is by FAR the short time you have to actually set up the shot, since this is not a controlled experiment where you know EXACTLY where the plane will fly and wait for him with the gun. It would still be a low probability shot even if you are ready for it, but not like winning the lottory.