Thanks, TU and Waz, your comments (and the link to Matt's post) really help.
For starters, it sounds like "Use Y" should be unticked unless you are working with a very large (i.e. planetary) scale, as you usually want constraints to follow the curvature of the planet. Yes? No?
And, just a simple example to help me get my head around this: Let's say you have a surface layer with a displacement offset of 100 meters. You connect the output of this layer to the main input of a second surface layer. Now, does this mean that a terrain/texture value of 0 for the second layer equals a final value of 100? And does this same relationship hold if the second layer is a child of the first?
I'm running into problems in deciding which setting to use for altitude constraints and fuzzy zones for overlapping layers. The scene I'm working on now uses several, and my initial thought was that I should set all of them to "Final position," so they use the same baseline and can be more easily coordinated. But the results haven't always been predictable. So now I'm wondering if I should use "texture/terrain" as the default, or use a mixture of both. But I'm not sure what to base the decisions on.