Sorry for the late update.
The idle mystery is solved… but it wasn’t obvious. While trying to get a handle on the poor idle control I noticed that battery voltage would start out fine but would sag. At start it would always be 14.5V, but a few minutes later it was down into the 12s. At the time I was concentrating on the idle and added it to the list of stuff to deal with later. However, idling the car for very long caused the battery voltage to get so low that it couldn’t crank, and a few people pointed out that the low voltage might be a component in the lousy idle, so attention turned to that.
Putting a multimeter on the alternator confirmed that the problem was real and not just the dash (that’s another story…). The possibilities were a heat-related failure of the alternator or a slipping belt, but neither seemed likely when idling for just a few minutes. Turns out there was a third choice. When I wiggled the wires plugged into the alternator, the voltage went up; bend the wires the other way and it dropped off again, so the problem was inside. Taking it apart and… well, I’ll have to yell at the guy who soldered the wires together.
With that fixed, suddenly everything behaved better – everything responded to reason! I suspect that the connection was being made and broken at about the same rate that the idle control was trying to correct it, and like pushing a kid on a swing, it would get worse and worse before causing the engine to quit. Cold start and idle are now vastly improved. Sometimes it still has an issue when approaching a stop though, especially if I don’t push in the clutch early. It’ll sort of bobble around at 800 rpm, then slowly recover back to 1000. It never stalls and it isn’t a big enough issue to stop working on other things.
Next was setting up open-loop boost control, using the same 3-way MAC valve that everyone else uses (and every company labels as their own). Somewhat surprisingly, I’m seeing the exact same high sensitivity to duty cycle as with the KPro. Examples seen elsewhere (including in the manual) show this same valve changing boost by 2 psi with a 10% change in duty cycle (five times less sensitive). I can’t figure out why my setup acts so different. I’m running a twin-scroll turbo so there’s two wastegates, but if anything I’d expect the increased air volume to make boost control less sensitive, not more. Maybe I have a “defective” boost control valve that works too well? Doubtful…
If the wiring to the boost solenoid was intermittent, it would have less control, that’s not happening. If there was a leak somewhere in the hoses or wastegates, it would have less control, not more. FWIW, early on (with KPro) someone convinced me that the entire problem was due to using the wrong valve, that I needed to use a certain 4-way MAC valve instead. It was even worse. Lastly, someone said I may have to add a small orifice (a restriction) in the hose leading to the wastegates, to in effect, desensitize the boost control valve. That should work, I’m just puzzled why it needs to be done at all. (I have a hard time believing that my boost control plumbing flows soooo much better than everyone else’s that I have to dumb it down.)
I haven’t engaged closed-loop boost control yet but wonder how well it controls the loop with such a sensitive valve.