It's quite amazing what can be done with logic chips, however I'm like you Ian, I'd be using software and I'm eager to see what Irving comes up with.
Most of the early Gecko drives relied alot of pretty basic Logic chips, yet Marriss managed to do some pretty cool things with them.

Assuming Dean doesn't require to move the exact number of steps/pulses, then you wouldn't have to be that accurate about timing, and could miss some pulses.
For example, as soon as a pulse is input, output x pulses at a slow rate, then if another input arrives before the x pulses are generated, reset the counter to x and increase the output frequency and so on. That way you're never really behind the curve in terms of number of output pulses, and you don't have to handle buffering/predicting what's going to happen. Off course slowing down is pretty easy, however you may want to implement some kind of deceleration so you're not slamming to a stop as soon as you stop spinning the MPG from high speed.

And from there, you could also add a logarithmic style input, in that the quicker you spin the MPG, the higher the step multiplier is i.e. slow turn could be single step, gradually increasing to 50x at high speed. That way you can still maintain accurate slow speed, yet still move at high speed.

I could probably throw some code together, however I don't have any MPG to try it, and I've also currently not got any spare Arduinos kicking around! (I keep meaning to order some, as I've got a couple more prototypes to build, but that's currently on the roundtuit list!)