Can still implement per-rail limits using a clever rectifier -- perhaps something that disconnects above threshold (so, doesn't dissipate power continuously) and resets every cycle, giving a characteristic similar to a peak current mode converter perhaps.
Might be easier to do on a forward converter also, in which case the primary side converter doesn't need control at all, that can all be done secondary side. Which also means no problems about cross-regulation, every output is individually regulated.
In the old days, converters of this sort were made using saturable reactors -- the reactor switches on after some pulse width, controlled by a feedback loop. You'll probably have a hard time finding reactors of the right value (flux and current), let alone in an affordable price -- but MOSFETs are more than good enough to take over.
Heh, I'm probably oversimplifying a bit, still. The control circuit would preferably be some IC that just drops in. I don't know any offhand, that are made specifically to implement this functionality. You might have to build your own from timers and comparators and such, which will end up taking up way more space (N controllers..).
Anyway, it doesn't have to be a full common array either, if it's adequate to have a say 300mA limit, total between a pair of channels, you can just use dual secondary transformers (assuming the secondaries have adequate isolation between each other) and primary side limiting. Now you need N/2 controllers, etc. Every little bit helps.
I did some more searching on transformer vendors websites and came across transformers intended for PoE. e.g. Pulse PA5099.003 or Coilcraft POE13P-24L. These are relative low cost, but 1:1.4-1.5 instead of 1:1.
So the whole thing might be easier, if i start with 36V and do 36V to 24V isolated converters. Or use the PoE transformers in reverse.
If those are flyback type transformers (relatively low magnetizing inductance), the exact voltages don't matter very much, you can use them higher or lower just fine. Or as forward converters, the extra ratio just as well helps provide adequate adjustable range (a half-wave forward converter loses half the secondary voltage due to PWM, so the required ratio tends to be higher than otherwise).
Tim