Is tonearm bias a compromise, maybe a myth?

Yes I agree with Lewm that AS should be applied, with due care.
In my experience as a dealer and 30 years of audio I have found that very few TT's provide a platter surface and armboard surface that are congruent, that is if the platter is level then the armboard isn't. How many of you have run dial calipers over the platter/armboard to measure this and corrected it.
If you dont run a unipivot or arm that includes adjustability for levelling the horizontal bearings, then you have in inward or outward force applied to the cartridge before you even start.
It is entirely logical to me that each arm may require different AS levels due to differing arm geometry, cartridge alignment preferences and bearing friction levels.

I think the most common issues with applying AS that I have seen are :
Compensating for an out of level arm relative to platter surface.
Compensating for output level imbalances between channels, either in the cartridge or system ( or even the audiophile's ears ! ).
Following the manufacturers recommendation without listening.

I can pass on my experience with the Ikeda MC which has no cantilever. The diamond is mounted on a hoop which is at a rightangle to the groove. The hoop is held by a string which prevents the hoop from being pulled out when playing. The hoop has a little compliance built in.
http://www.avguide.com/blog/high-end-audio-japan-part-five-analog-ikeda-sound-lab
This cartridge requires AS on every tonearm I have tried it on, both unipivots and gimbal bearings. No antiscate the you get mistracking, channel imbalance and compressed soundstage and harmonic structure. Optimum AS applied yields no mistracking, no channel imbalance and most importantly maximum soundstage and harmonic structure.

Lew,

I agree with the elaborations in your post. For clarity, the only point of my (admittedly over-simplified) analogy was to illustrate this fact:
- skating forces act (first) upon the stylus
- anti-skating devices act (first) upon the tonearm

This seems non-controversial, obvious even, but this difference must be mediated somewhere. Practically speaking, the only non-rigid component between the stylus and the tonearm is the suspension in the cartridge. That's where imbalances between skating and anti-skating forces will be dampened. To the extent such dampening occurs, sonics will be affected.

Skating is a component of the force vector generated by the friction between the stylus and the vinyl. If the stylus on a pivoting tonearm is not in line with the center of the pivot bearing, then the x and y components of the friction force try to pull the stylus out of the tonearm, ie. away from the bearing and also left (towards center) of the tonearm because the stylus is to the left of the pivot bearing. Linear tracking tonearms have no anti-skating adjustment because the stylus is in-line with the sliding arm tube bearing. It has nothing to do with the grooves. The tonearm will be pulled to the center of the record even with a blank vinyl disc. Since the counter force, or anti-skating force must be applied at the pivot of the tonearm, cartridge manufacturers apply a bias or prelaod to one side of the stylus/cantilever suspension just like they apply a preload to the suspension in the vertical direction. That is so when the cartridge is cued onto the record, the VTF load and AS load center the stylus in its suspension- as long as those loads are to design intent. Try to look at the stylus cantilever on your cartridge sometime when it is in the free state. It should look offset down and left. Then cue it onto a record and it should look centered and left. Now start the record turning and it should look centered. That is if the loads are adjusted within the range of the cartridge manufactureres' specifications.

Tony, interesting post. Regarding pre-loading and non-pivoted arms : does this mean that the cantilever will be permananently skewed when a cart is fitted on linear tracking arms?

I can't say for certain, but I think that the stylus/cantilever pulls the tonearm along, so if those loads are balanced properly then it is going to be centered. I had a buddy who used the ET2 linear arm on his VPI table. It took him a few years to figure it out, but I recall him telling me that he had to level his table a certain way- slightly downwards, I think and it would track perfectly. He also said that he had to keep the air bearing tube perfectly clean. It would sound grainy if the tube had any dust on it.
Also, cartridges are different. I think some may have a different compliance in the horizontal direction than in the vertical direction, ie. the cantilever may be stiffer in the horizontal direction. So you may not see as much deflection horizontally on some cartridges. My Benz cartridge has what looks like an o-ring on the cantilever for a suspension. So the bias downward and slightly to one side is very apparent. But when riding on the vinyl it is centered. The cantilever goes through a hole in the housing which makes a good reference for seeing the cantilever position.
When I am setting up my tonearm and setting the HTA, I have the protractor on top of a record so that the arm is at the right height. Anti-skating is set to zero. I pull on the platter while viewing the protractor alignment. This applies friction drag to the stylus and sets the cantilever in the "as playing" condition. I can see the difference in the protractor alignment when turning the record slowly versus the record remaining stationary. (The record slides along under the protractor).