Interesting and rather unexpected result. The lines off the right side suggest that the timing of the HSYNC pulse, rather than the length of the retrace period, is the problem. It looks like retrace is starting too late so it isn't finished by the time the next video line starts.
I might start with the base driver circuitry for Q2:
When Q2 is on, charge is stored in its base which needs to be pulled out for Q2 to turn off. That's the function of Q1 and transformer T1. If for whatever reason this circuit was limping along at half-power, Q2 would take too long to turn off and the retrace would start late. The signal at the base of Q2 should look something like this:
The vertical scale is small, 1V/div. During retrace you should see a negative pulse of -2V or so. The rest of the time it should be positive, around .6V or .7V (when Q2 is on). Turning on is gradual (sloping line), where turning off is fast. This plot was taken with just a resistor as load on Q2 rather than the yoke, so I'm not sure yours will look exactly the same, but it may be similar. If the turn-off is slow or the negative pulse is particularly narrow I might be suspicious.
On the other hand you could just replace Q1 and see if it helps.
Check around this area and make sure there are no other obvious problems like bad solder joints, and double-check the resistance of R17. It's possible T1 has failed but you won't find a replacement without a parts donor analog board.
You could also try comparing signals on P4 pin 10, U1 pin 3 and the collector of Q1. They should all have approximately the same signal (HSYNC), variously inverted and level-shifted.
The odd part of the scan lines picture, though, is that the scan range is shifted off to the right on average (since it goes further to the right than the left). Did you manually recenter the image when you started? If not, that would mean there's a DC current through the yoke pushing it off centre. C15 is supposed to block DC current, so I'm not sure how that would happen.