Hmmm! A number of issues here. First, I don't believe that adding a VFD will reduce the code requirements for conductor and fuse sizing. Second, the cost comparison between a RVSS and VFD is related to HP. Under 10hp they are nearly the same price. By 100hp (both of these are assuming 480VAC) the ratio is about 4/1 with the VFD being more expensive, of course. Third, about the best a RVSS can do even on a free shaft zero starting torque load is to cut the inrush by 60%. There will always be some starting inrush. On the other hand, since a VFD starts a motor nearly synchronously, there is no inrush as long as the starting torque does not exceed the nameplate torque. So a VFD does a much better job of reducing inrush while preserving starting torque.
Just for reference, when you reduce the starting voltage on an induction motor, the inrush current comes down about in direct proportion to the voltage reduction. But, and here is the problem, the starting torque comes down as the square of the voltage reduction. So, by the time you reduce your inrush by one-half, you only have one-quarter of the torque left. By one-third, it's down to one-ninth and the motor probably won't start itself, much less any connected load.