My MJ did the same thing.....I replaced the Manifold Air Temp sensor, added an extra ground from the battery to the front sheetmetal, and cleaned all the other grounds really well. Don't know which one fixed it, but now it works fine........
Given the way RENIX is built, I'm inclined to think it's a combination of cleaning up the grounds and improving the ground for the chassis.
OEM, the chassis grounds through the engine block, which is silly (and one area where AMC screwed up.) That's why I offer the additional chassis ground as an option on my RENIX kits - and I've gotten a lot of good reports from the field on it.
I do also suggest that the contact area around the firewall ground strap be cleaned down to
bare metal and a corrosion inhibitor (you can find something in your hardware store's electrical section, I like Gardner-Bender Ox-Gard, myself...) on assembly. Clean the lugs as required with a wire brush or some very fine (320-grit or finer) sandpaper. The objective isn't to remove material, just remove crud.
@OP - this is probably what I'd do next. You're saying you've verified operation of the TPS (did you use an analogue meter, or a digital? You can backprobe and check voltage KOEO or KOER, or remove the connexion and check resistance KOEO.) This sounds for all the world like my 88 did when it developed a "flat spot" around 2300-2600rpm, and replace the TPS nailed it.
I've been kicking this around, and it seems as though you've gotten all of the other likely suspects.
The reason RENIX benefits from an improved ground return is simple - the ECU is mounted inside the cabin, and the ground reference plane is the chassis sheetmetal. If the ground starts to get stupid on you (at the firewall strap or at the engine connexion,) your sensor signals will start to be corrupted by the floating ground. Adding a ground directly from the battery to the chassis (treat as for the firewall ground strap, above) helps to prevent this as an issue.