Well, I once went with full disks in 20 mph winds (and open field) just to test how it works:
You do have to lean into the wind, but otherwise only time I've felt *something* approaching 'wind steer' is when I've been hit by an amplified gust of wind channelled between two buildings (Venturi effect), and I can swear than it was *out* of the wind, not into the wind, despite having about 60mm of *negative* offset. It seems that even 60mm of negative offset is not enough to, ehehe, offset the lateral area of your legs, crankset and other bits that in front of steering axis (at least in my case).
What is *worst* with positive offset, however, that it leads to what I call 'negative aerodynamic trail' or 'windvane' effect - steering corrections at speed would generate torque 'into the turn' - which is opposite of what is required for stable steering. So, in theory, at high enough speed, negative aerodynamic trail (quadratic with speed) may overcome positive trail that is based on friction and rises linearly with speed and make the bike behave like one with negative trail. Very, very scary thought.