Assume you make a glider which is a big cylinder with a huge
hollow tube going through the middle. I lay it on its
side and put a clear plexiglass on the top part for the pilot to
see out of, and give him a seat inside. Then I put three yawstrings
on it: one on the plexiglass "canopy", one in the middle of the
center tube, and one on the bottom.

I drop this "glider bomb" and it heads straight down (maybe there's
a drogue chute). A gyro rotates the cylinder on the way down.
Assuming no surface friction and ignoring gyroscopic
precession for now, all three yaw strings, from the
pilot seat, show different things. If the cylinder is
rolling right, the "glider bomb pilot" sees the yawstring on
the canopy and instinctively wants to add right rudder.
The string in the center of pressure shows straight, and the
bottom string would make the pilot want to add left rudder.

None of this has anything to do with gravity, adverse yaw,
or the cylinder slipping or skidding.

I contest that there is an error caused on the yaw string
depending on the roll rate, airspeed, and the distance of the
yaw string above the center of pressure, and this will
always tell the pilot to add more rudder in the direction
of roll (assuming the yawstring is above the center of
pressure), i.e. skid.

The size and importance of this error is another
matter entirely :PPP

Well, the size and importance DOES matter because you are making such
a point of it. I think that I adequately proved that your last point
was, shall we say, pointless. The offset of the yaw string to the
center of the roll axis is much less than its displacement from the
CG, making your "errors" proportionally less. I think you ought to
pursue a more reasonable hypothesis; like TWA 800 was downed by a
stray Navy missle.

Tom