I guess I'm 6 years late to this topic, but have some info to add. First, the idea of using diodes in series with the motor will limit maximum speed, but it will also make the throttle knob unresponsive until it is turned up enough for voltage to overcome the drop across those diodes. For that reason, I did some testing with an AC transformer at track different voltages.
Attached is a chart of maximum speeds (in scale miles per hour) for the Diesel from a 6-84719 Santa Fe Super Chief set and an 0-8-0 Steam Engine from a 6-84725 Southern Pacific Rising Sun set. They are somewhat similar behavior from 14 VAC to 18 VAC with perhaps 16 VAC being the "sweet spot" for a nice top speed of 60 mph.
Also, I did current draw tests under different conditions for the two engines. Running at a moderate speed, the Diesel was 300 mA and the Steam engine 150 mA. Stalled at a moderate speed setting the Diesel was 700 mA and the Steam engine 450 mA. Stalled at the maximum speed setting the Diesel was 1500 mA and the Steam engine 1620 mA. Since the 36 watt "Wallpack" that comes with each set is 2 amps, it can easily handle both trains unless both were stalled at maximum speed settings.
The series diode scheme, instead of in the engine, might work fine in series with the Wallpack's fixed DC voltage to drop the track voltage. The Wallpack is DC so only one string of diodes would be needed (unlike with the motor where two strings are needed - for each direction). It would be best to use 3 amp diodes minimum. With trial and error the number of diodes desired could be determined and then hardwired. But a scheme could be set up using barrier block terminal strips with a diode between positions 1 and 2, another diode between 2 and 3, and so forth for as many possible voltage drops you might ever want. Then you can always have a number of places to connect the power wire to the track at different possible voltages each 0.6 VDC higher than the next.