Try looking at it this way. Take a 1/2" sq stick of soft balsa 3 feet long and draw lines on it every inch ina direction perpendicular to the length. Now bend it into a curve. You'll discover the lines you drew are still straight BUT no longer parallel. Instead, they now all point toward the center of the radius of curvature. Therefore on the outside surface of the stick the material has stretched outward due to tension and the inner surface of the stick has compressed.

Axial stress occurs when you try to pull on the ends of a bolt. Shear stress occurs when you use the bolt to clamp two plates together and then try pulling the plates apart. The bolt is said to be loaded "in shear". Thats why big scissors are called "shears"

Back to the 1/2" stick. If the outer surface is stretched and the inner surface is compressed, then we have internal shear. The shear is tangent to the arc of curvature in pure bending. No g-loads, webs or balsa heterogeneity to clutter up the concept either.

What Rodney failed to understand in all of his study in this area is that the shear in a cantilevered beam varies from zero at the tip all the way up to the full bending moment divided by the second moment of inertia at the root yet the vertical load is constant. Therefore at some point in the wing the shear will be at 45 degrees but it's different everywhere else. Furthermore, the point of maximum stress in web shear is dominated by the pure bending shear at the root where the worst case occurs. Shear in pure bending is horizontal as illustrated at the beginning with the 1/2" stick example.