Well, after a long break I finally got back at tuning the Holley. Last year I already installed some bowl sight plugs and had adjusted the float levels a bit below the bottom of the hole (which is the normal rule of thumb setting). But that didn't make any difference.
This time, I took Drew's advice and popped out the idle well plugs. Everything was nice and clean in there, but these carbs have the idle feed restrictors hidden in this well. Measuring their size with a 1 mm drill showed that they were probably ~0.041". I had no idea if that was big or small, but some googling showed that a normal size for a ~750 cfm Holley is in the low .030 range. In combination with my mild roller cam (and thus extra high vacuum), this was probably the reason for the rich transition circuit.
I stuck in some small wires and made some new caps out of aluminum (I didn't have any brass ones). That leaned out the transition circuit somewhat from 10 to 11.5 and thus not nearly enough. At cruising speeds, the AFR creep up somewhat but dropped back to 12 right away due to the power valve. Looking at the vacuum gauge during driving, I concluded that the 8,5" power valve kicked in way too soon. I then installed a 6,5" PV which allowed the AFR creep up to 14, but the power valve still kicked in too soon, especially in fifth gear.
I took it apart again, inserted a thicker wire in the idle feed restriction (0.024", reducing the flow area by 1/3th) and at the same time, I also installed #66 jets (instead of the stock #68), a 3.5" power valve and the stiffest (black) spring in the secondary canister (it had the next to lightest yellow one, which I assume is what it came with). Not very scientific perhaps to make so many changes at the same time, but I had grown tired of taking the carb apart all the time.
Anyway, idle AFR remained at ~13, light cruise (on the transfer circuit) increased the AFR to ~13, cruise (on the main jets) to 14.5-15 and WOT to ~13. It smelled much less and drove smoothly without much hesitation. There's still some under some conditions, so a little more tuning of the acceleration circuit required I guess, but overall it drives well. At 6000 RPM, manifold vacuum crept up to maybe 1" Hg (hard to see exactly when trying to look at the tach, the vacuum gauge and of course the road at the same time...), so I guess the stiff secondary spring didn't make the carb very restrictive. The light cruise AFR didn't increase as much as the flow area reduction suggested but I guess that's due to the unchanged secondary idle circuit. Maybe I'll give it one more go and fix that too.