place to start would be to design a competent push-pull tube amplifier and then disable the drive signal to one leg, unless the signal is high enough of a value to imply that the other leg is nearing cutoff.
You have in the Dynaco MK 3 a fairly good push-pull. Why not forgo the trendy aspect of the para-feed for the ease and simplicity of just converting the MK 3 into an SE / push-pull amplifier by retaining the basic layout and functioning of the stock amplifier? To see what I am suggesting, imagine if you were to remove one of the coupling capacitors in the amplifier.
One tube would still see the signal and would amplify that signal. But the other tube would only see the negative bias voltage and not amplify any signal. Thus, we would have half an amplifier. The output transformer will not saturate, as the disconnected tube will still draw the same idle current as before, which will match the idle current of the other tube and prevent any net magnetizing of the output transformer's core. The amplifier would then work in SE Class A mode, as just one device handles all of the output waveform.
Now if we replace the capacitor only during that part of the drive signal when the normally operated tube ceases to conduct, the amplifier becomes a push-pull amplifier. (One point worth noting is that the impedance presented to each tube is just 1/4 the impedance the same output transformer would present to two tubes working Class A push-pull. Why? One half of the primary winding is used to couple signal to the speaker at any time. So the winding ratio has now dropped by half and the winding ratio squared times the load impedance is the reflected impedance and [1/2]² = 1/4.)
The amplifier redesigned should meet the requirements for SE/ push-pull operation. The first change to note is the move to the pentode configuration rather than the ultra-linear of the output stage tubes. This was done to keep the normally idle tube idle, as the signal from the ultra-linear tap would stimulate it in the same way as signal at grid 1 would. The second change to note is the insertion of a choke at the idle tube's cathode. To better understand what is going on here, remove the diodes, zener and rectifier. Now imagine what the cathode signal