Signal injection to the rescue! Used a B&K 1077B to isolate the problem to the 3rd IF stage.
After checking resistors, coils, voltages, etc, I rechecked the tubes. The 6BA6 at this stage had 0 emissions, no filament glow and got very hot in my tester. Must have lost its vacuum. That's a shame as it was an industrial, long-life version.
This is one I had already replaced after my first round of testing.
I replaced it with a tube from my stash, but no improvement. I rechecked the replacement tube and it was bad too!
So, replaced it yet again with a NOS RCA, bingo! We have reception and sound!
I can also now see that the yoke is on wrong as the image is upside down and backwards. Wide bars should be on the left.
That means the horizontal scanning issue is actually on the flyback side, not the drive side.
I'll work on that next.