The airloads will be too large to put into the servo body. Remember that the servo body has to respond to torque alone. The tail inputs a really really large bending moment into the output shaft. It would break pretty easily. So you need linkages to handle the up and down loads on the tail.

Slats, flaps, etc will all be fine and aerodynamically sound if built correctly.

You asked, "But, again, the question I guess is to what degree control surfaces on full scale a/c can be usefully replicated with smaller versions at slower speeds?"

Exactly - unless you are building a yard flier. For example, in full scale F-15 design work we used a 5 to 6% scale model for anything Mach .5 or above. For low speed work, landing, etc. we used up to a 12 to13% scale. As I set here the exact numbers escape me (getting old) but are certainly representative. The particular sized used were dependent on the size of the wind tunnels that were available to test in, not the fear of not getting representative, good quality data.

The thing is that we were absolutely certain of getting aerodynamic answers that were right on and that were sufficient to convince a stingy (rightfully so) government to give us money.

With that in mind it is very reasonable to go to much lower sizes of models and still get something close to full scale accuracy.

Bob Violetts F-100 and the others mentioned above are good examples of airplane size and flying qualities.

You should feel totally condifent using the scale control surfaces at the sizes you are building. Maybe not if you are building a slow flyer for the backyard.