Although I'm very happy with the results, there's clearly still some downward slope to the current before it falls off the cliff. I'm not sure where that slope is coming from. I would need to monitor the voltage at the current limit input pin to know for sure if the current limit is being held constant at the PWM controller, or if there's something else going on.
I'm also not sure about whether the last two mods have any effect. I strongly suspect that supplying the correct AC voltage gave most of the improvement. The last 2 mods were an attempt to solve the problem that I now attribute mostly to low input AC voltage.
When I'm back from vacation, perhaps I'll have a chance to understand the circuit a bit better.