I have no experience with these (andnot too much with switching power supplies in general) but I found a schematic/manual here: http://home.earthlink.net/~gamba2/images/plus_analog.PDF

I'd check for a switching signal from U3 first. If there is none, the fault lies somewhere in the feedback circuit. Check for a signal on pin 7 of U1 in that case.
An easy way to test the zener is simply to check for 6.2V across it.

(also how is U1 supposed to monitor the 12V rail properly if it's powered by the 12V rail itself??)

As long as power supply is large enough for the circuit to work (about 7.7 volts or greater), U1 can compare the divided-down supply voltage to the 6.2 volt reference so the feedback loop tries to control the supply voltage until both op-amp inputs have equal voltage of 6.2 volts (the voltage divider network is adjusted with a trimmer so that 12 volts gets divided down to 6.2 volts).

I've pulled the optocoupler and scoped the output of the op-amp driving it, there's a nice rail-to-rail pulse width modulated signal controlled by R56 and also the 6.2V reference is OK so I assume the feedback control circuitry is good and my problem lies on the other side of the optocoupler.

Replaced the optocoupler with a 4N26 and no change, I'm suspicious of Q10 (30V 0.8A fast recovery SCR) but I don't have any of those at hand, just a boring slow TYN412.

EDIT: Stuck Q10 on a protoboard with a LED and found out it's leaky, so I believe I found the problem. Now I gotta wait until I can get a replacement.

Got a 2N5064 as a replacement for Q10, along with finding that the original optocoupler was also bad and R44 (100Ω) had drifted up to 133Ω. Replaced all three and now R56 does have an effect on the output voltage, but at full line voltage the OCP/OVP continually kicks in even at the lowest setting. Once again slowly bringing up the AC through the variac I can get the same wavy raster as before, as soon as I get close to normal operating voltages OCP/OVP fires again, but with some careful twiddling I was able to get a stable raster and a successful boot, accompanied by that infamous sizzling sound of frying bacon that is so characteristic of a high voltage leak, and sure enough, using myself as a capacitive return path this happens:

It leaks all over the place and a few seconds later we're back into overload. My conclusion is that the flyback is partially shorted, causing it to draw too much current in normal conditions, but at reduced voltage/current it sort of works, enough to produce some raster at least.

It's clear I'll be needing a replacement flyback, but there are a few other things that bug me. With the flyback completely removed the power supply still overloads like before, even when putting some dummy load on the +12VF line which makes me believe there's still something wrong with the feedback/regulation circuits. I remember a while ago when repairing a friend's Mac Plus (international version) that the PSU would run fine without the flyback, which makes me suspicious that this one doesn't. Another thing is that The Dead Mac Scrolls and the official Apple schematic calls for 1N4001's for CR28 to CR30, while Classic Mac Repair Notes call for 1N493x. Given that the PSU oscillates at tens of kHz the fast recovery diodes make much more sense than simple 1N400x's.

While I won't be able to do much until I can get my hands on a replacement flyback transformer, I still think that something's not quite right with the power supply, maybe the choice of replacement opto/SCR could have to do with it, or it could be something else I'm overlooking.

Have you checked/replaced capacitors? Ripple is characteristic of a bad cap

Did a cursory ESR check on all of the big caps and they were fine, replaced C1, C2, C35, C36, C9 and C43 anyway. Doesn't mean the rest couldn't have reduced capacity and/or leaks, and I'll be replacing them anyway, but in this case the ripple was caused by reduced line voltage. I was able to get a ripple-free raster for a bit but then the flyback started sizzling and leaking HV.