Nsgarch is correct on all counts (except the mud). Here's an example I often use which seems to help people visualize the stylus azimuth conditions he described.
Sit in a desk chair or rocking chair that can tilt fore and aft. Face directly forward. Extend one leg out in front so your lower leg descends to the floor at an acute angle, with your heel resting on the floor.
Holding your leg in that position, rock the chair back and forth while watching the arc described by your heel. If you're facing directly forward (perpendicular to the chair's axis of rotation) the heel moves straight vertically up into the air and back down.
Your heel is the stylus, your lower leg is the cantilever, the chair is the tonearm. When the tonearm's axis of vertical rotation is perpendicular to the cantilever, vertical arm rotations (as over warps) do not change the azimuth of the stylus.
Now scootch yourself around so you're sitting at a 20-25 degree angle to the chair. Extend your leg again and rock the chair as before. Observe the change in angle between your heel (stylus) and the ground (record). The sides of your "stylus" no longer remain vertical.
When the arm's axis of rotation is not perpendicular to the cantilever, the stylus moves through an arc and its azimuth changes relative to the record. This is the case on all tonearms with offset headshells whose vertical bearings are not offset at the same angle.
When azimuth alters with every change between record surface height and arm height (as over warps or with different VTA settings) the results will include variable crosstalk, muddy imaging and potentially uneven record and stylus wear.
Why build an arm with the vertical bearing axis perpendicular to the armtube rather than the cantilever? Because it's cheaper to design and machine at right angles. The design is inherently flawed, but when building to a price point manufacturers must make compromises. This is one they sometimes choose.
