UPDATE: We pulled the engine and trans today, and put the correct flexplate on, which definitely has a different geometry and sits closer to the trans (i.e. not pulling the torque converter out). There was no evidence of spline or seal damage. Put it all back together and ... it runs, drives, and shifts perfectly. So my theory, #1 below, was correct. Also changed real main seal and torque converter seal since it was accessible.
Moral of the story: Always use your original flexplate / ring gear. Make sure the torque converter is seated properly, and don't pull it out if the studs seem too short. :)
So, I talked to 3 friends about my problem, and got their ideas on what may be wrong. One of them actually helped me swap the engine (extra hands), and the other two are mechanics who work in the same shop where I did the work (but weren't directly involved in the project). Based on my description of what I did, it turns out they all have the same suspicion:
- I probably pulled the torque converter out a bit too far when bolting it up to the ring gear. The output spline is long (~3/4 in), but the spline that drives the transmission fluid pump is very short (~1/4 in). Pulling the torque converter out too far would cause this pump spline to slip, which means no fluid pressure in the trans and no engagement of gears.
There is some evidence to support this theory: My torque converter has 4 studs that protrude thru bolt holes on the ring gear, and nuts are attached on the back side of the ring gear thru the starter opening. When we first connected the new engine to the trans, things slid together very nicely, but oddly the bolts did not protrude thru the openings in the ring gear, although I could see them flush just inside the holes. I put my fingers around the edge of the ring gear and "wiggled" the torque converter closer, until the bolts began to protrude, then attached nuts and gradually tightened them in sequence. I naively thought that the torque converter had simply been "pushed back" during the assembly of engine to trans; I did't have enough experience to realize that is probably not possible. If this theory is correct, it means the ring gear on the new engine is different than the old ring gear, or some other geometry is out. I know I should have used my original ring gear, but a visual inspection led me to believe they were the same!
When bending the wiring harness out of the way, I could have broken some wires that connect the ECU to the trans. I feel this is unlikely, because I was pretty careful not to stress the wiring too much, and there is no evidence of hard or cracking insulation even on the wires that are exposed.
Some sort of contamination could have entered the trans fluid lines between the removal and re-installation of the radiator, which includes a trans fluid cooler, and fouled up the valve body or some such. (I'm pretty sure there isn't a leak, because there is no evidence of coolant contamination or loss of coolant.) But maybe some debris got into one of the hoses when it was disconnected. But I feel this is unlikely, even though I didn't cover (cap) the hoses and connectors, the area of the shop I was using was undisturbed during the three days I was there.
Air that entered the trans cooling lines could have somehow fouled things up. I think this is unlikely, or it would be a problem every time somebody changes a trans cooler / radiator.
The transmission, although high mileage, drove just fine before the swap, and just coincidentally failed while I was working on the engine. How likely is that?
I'm going with #1 for now. Unfortunately this means I need to pull either the engine or the trans again and put my original ring gear on. I will also make a careful comparison (measurements) of them. My friends feel that it is unlikely that I damaged the trans in all this, so I'm hoping they're right.