As expected, disconnecting the pony run cap did the trick. Mystery solved. Everything works perfectly, and even better, I understand why the problem occurred in the first place.
One curiosity that someone raised was how much the pony heating was mechanically loading the idler. Switching out the capacitor reduces the line current into the electrically unloaded idler from 8.4A to 5.2A. Assuming the idler's power factor correction is perfect, that works out to about 768W, or a tad over the pony motor's 1HP rating. So, the pony was working about as hard as it could to slow down the idler without burning itself up.
Actually, I had used a variable sheave on the pony motor and adjusted the diameter to match the fixed idler sheave. Since both motors run at the same nominal speed, the pony wasn't being overdriven by much, just whatever the difference in slip is. That's probably why I avoided the "funeral pyre."
Thanks, Jim, Jon, and everyone else who weighed in on this. It was an interesting excursion, and I'm jazzed with the results!
Cheers!
Jim