Six weeks have passed. I hesitate to say that the effort has been successful, for fear of bringing the wrath of the God of the Vengeful Armadillo down upon my garden again, but plugging up that hole seems to have done the trick. It's just like old times ('old times' shall mean the time since Dennis built the fence for me in fall '05).
Six weeks without an armadillo is positively correlated with plugging up a second hole in the fence, and highly statistically significant. Plugging up the second hole under the fence is also practically significant, as my thriving garden demonstrates.
Plugging up one hole but not the other was totally ineffective to stop armadillo adventures as the two variables are completely unrelated. Fussing about the cliff and hypothesizing how an armadillo could scale it was also a big waste of time, uh, I mean, an unrelated variable, so long as a second hole under the fence remained accessible. But, if the first hole had not been found and plugged (the one just below the level of the first cliff edge), and the second hole had been found and plugged (up at the level of the garden proper), then the fussing about the cliff and hypothesizing how an armadillo could scale it would not have been a big waste of time, but plugging up a hole under the fence is a lot easier than building up a cliff with lots of heavy rocks to make scaling it difficult to impossible for an armadillo.
So, lessons learned: find the damned holes. Check really carefully. The fence does its job so long as it goes down to the ground and just a bit below and doesn't suggest to an armadillo going along its edge that there's a spot here that looks a little weak, and let's see what's on the other side...