I think there's a way around (no pun intended) calculating the radius that worked for me. Maybe I just got lucky, but this might spare you a little time on the test...
You know that the acceleration downward is always 0.4g =~4m/ss. Over the course of such a fall, the average speed should then be 2m/s. So every second, you need to move horizontally fast enough that you have 2 extra meters of space below you.
The curvature is already nicely provided: 3.6 km for every 2m drop-off. Therefore, (c) 3.6 km/s.
I think v^2/r is built into this somewhere, but when I was screaming through this test and timing myself, it seemed a little more intuitive and economical to me.