Well, not the same reason. Shear strength is in many ways more important because failure usually propagates from an origin like a rip- by shearing.
But fundamentally, tensile strength and shear strength are both much more empirical than compressive strength. Tensile strength can be used with ~99% accuracy on a wire that is microns long or miles long. If you double the relative width of a testing sample it will dramatically change the measured compressive strength even without buckling.
This is why the code forces you to use strut and tie model deep concrete beams. As much as you may want to idealize a shear stress, really what's happening is the beam is arching over the span.
"All failure is shear failure" - this is a simple explanation of Tresca's Yield Criterion. For materials with higher compressive than tensile strength, the equivalent is the Mohr-Coulomb failure criterion.
This is true, but neither is "tensile strength" really a thing for the same reason. A simple uni-axial tensile test is not really uni-axial, but a combination of orthogonal normal stresses that ultimately results in shear failure. I've heard it said that "all failure is shear failure", and I think that's true. When you look closely at the ductile fracture surface of a ruptured tensile specimen, the characteristic "cupping"[0] appearance consists of various surfaces at 45 degrees from the direction of the applied load. Principle shear stresses are always oriented 45 degrees from the principle normal stresses.
[0]https://upload.wikimedia.org/wikipedia/commons/1/1b/DuctileF...