Nice. The component sensitivity problem with S-Ks is very real if you need accuracy at even moderate Q--with a Q of 10, the logarithmic derivative of the centre frequency with respect to the component values is something like 30.
It's also true that you can do good things with automatic tweaking, especially of offsets.
I'm working on a fancy laser noise canceller that has about 7 automatic tweaks for both AC and DC, so that it can adjust all the errors out to order (omega)**2. It's a lot of work, though, and it you really have to pick physics-based tweaks as opposed to just fitting a curve and shoving a correction current in someplace. Otherwise the accuracy falls apart when conditions change slightly.
Automatic gain tweaking is harder, though--you have to mux between the signal input and a voltage reference, which requires a DPST switch per channel, and maybe two of them if you want to reduce crosstalk from the disconnected input. The hardness of this is part of what makes the S-K's excellent gain stability such a win.
Which design is superior depends on the design constraints.
Cheers
Phil Hobbs