"Straight" talk

lewm

... on average, one might be better off with ignoring azimuth adjustment in favor of setting the cantilever 90 degrees with respect to the LP surface.

I absolutely agree. Moreover - although every aspect of phono cartridge alignment geometry is important - I do not think that azimuth is the most critical. Much, much more important is assuring that the cantilever/stylus is truly tangent at the chosen alignment's null points. Gauges that align the cartridge rather than the cantilever are particularly vulnerable to getting this wrong. That's why I always suggest the use of a mirrored gauge, such as the WallyTractor or Mint gauges. These use the reflection of the cantilever itself to ensure proper tangency. The only better way, imo, is to use instruments, such as an oscilloscope.

I've been all around the block with azimuth, aided and abetted by the fact that I own a Triplanar, which was once one of a very few tonearms that allowed precise adjustment of azimuth (albeit while messing up zenith).  (Nowadays, there are many tonearms or just headshells that offer this feature.)

Plus, I own a Signet Cartridge Analyzer, the grandfather of the Foz.  While obsessing over crosstalk using the Signet and the Triplanar, I found I was occasionally messing up the sound by canting the cartridge extremely to one side or the other in order to make the meter on the Signet happy.  This was the case in particular with my Koetsu Urushi, which was canted way over, like folkfreak said, in order to achieve best crosstalk numbers.  When I just gave up and set the Urushi so the cantilever was perpendicular, the sound was and is a lot better.  Maybe crosstalk isn't optimal, but all else is copacetic.  Now I have become a nihilist where azimuth is concerned.  I think a lot of this has to do with the perfection or lack thereof with which the coils of the cartridge are positioned inside the body, which is something one can not directly observe.  If the build was flawed, then pushing the azimuth to fix the problem is not without its cost to overall sound quality (and probably aberrant stylus wear).  I don't claim that it is not a good idea to have crosstalk maximally adjusted when possible.

Re results that end up canted. As I’ve posted before while I own a fozgometer I much prefer (and get better results from) aligning azimuth by ear using this method. 
http://www.durand-tonearms.com/Support/Support/azimuth.html

Folkfreak, At first reading, I find a few problems with the blurb on the Durand website.  First, correct azimuth is ideally as they say, with the stylus perfectly centered in the grooves (however vague that is).  But most cartridges are not perfectly constructed, and the issue of crosstalk is most related to how the cartridge "reads" the L and R channels.  Reading takes place at the other end of the cantilever, where the coil moves between stationary magnets, in the case of an MC type, or where the magnets move between stationary coils, in the case of an MM type.  If the interface between the stationary and moving parts is not perfectly symmetrical, that will cause unequal amounts of crosstalk.  Second, they imply that badly adjusted azimuth is a major cause of channel balance problems.  That is simply not the case, and you can prove it to yourself by playing with the azimuth adjustment.  I found that extremes of azimuth adjustment affect channel balance by about 1 db at most, using the Signet analyzer.  Moreover, most texts on this subject will agree with what I say.  Playing with azimuth is no way to treat channel imbalance.  Azimuth affects crosstalk. I recommend treatises by Victor Khomenko of BAT and by another guy whose name escapes me.  If you go on Vinyl Asylum and search on "azimuth", I think you'll find their lengthy posts on the subject.