Absolutely motivated by cost/availability of power PNPs and its influence on performance tends to be negative, that's why the designers bothered with that relatively complex output stage with three transistors per side and local feedback.
Voltage across 0.33Ω resistors definitely shouldn't be zero, that means no bias at all - you know the Ohm's law, I hope. Adjust R609 for the same voltage across resistors as on the other channel and be careful because it may increase or decrease after warmup if thermal compensation isn't perfect. Alternatively, if you can find complete service manual, it should tell how much bias to set. And you could test those resistors if their value is right.
edit
Now that I think about it, if both channels are set for the same bias and run at the same temperature when idle, but one gets hotter under load, then C604 may potentially be the culprit. It could result in temporary increase of D604,D605 current during negative output cycles and some corresponding increase in bias, maybe(?). I'm not entirely sure, but it's a cheap and easy component to replace...