The problem I see with that is that the base current correction only works at DC. For few-percent error tolerance, it's probably fine, but the step response will have whoop-de-doos. (They may be too far out in time to care about, but they could also interact with subsequent pulses and cause jitter.)
I still say that simulated inductor circuit is simpler--4 parts, zero current error at all frequencies, L as high as you like (if you use a BFP640). You can put a damping resistor in parallel to kill the Q if that's an issue. (I suppose the whoop-de-do jitter issue might be there as well, somewhere between the bandwidths of the LC and the op amp loop, but I'd have to think way too hard to figure that out.) ;)
Cheers
Phil Hobbs