The biggest advantage of very high altitude reducing the friction problem isn't just that it stops your projectile from turning into plasma, it is that the size of the minimum viable projectile is reduced. People keep focusing on getting people or vehicles into orbit, but the main advantage of a gun-to-orbit system is that the packet size can be made very small. Think machine gun instead of cannon. This means a smaller gun and thus smaller capital investment.

Instead of the balloon-supported long gun that mikewarot suggested, imagine a balloon at 30 km altitude supporting a 5 meter long gun and a few tons of both propellant and <1 gram pellets. A target satellite is shot in a long continuous burst, leaving the satellite with a lot more mass and a slightly disturbed orbit. Later shots from a different location and direction can be used to compensate for the added impulse to the satellite's orbit.

This could be a cheap way to get a bunch of metal into orbit, or possibly even fuel and oxidizer.

Projectiles of what mass? And with what acceleration? As I understand it, with feasible railgun lengths the acceleration is well above the limit of human tolerance, and the projectile mass is significantly smaller than a typical payload put into LEO by conventional rockets.