I do see your reasoning, but the current available to charge that cap will be 23.5ma, 90ma won't pass through the top CCS. Also the cathode voltage and anode voltage are both zero until the shunt regulator starts conduction, so the reference terminal is free to vary a bit. This has worked on the shunt regulators in many Extended Foreplay III's. If we had a fet in there instead of a tube, then I might be pretty worried.
Good point on the current source. How much current does the tube sink? Assuming worst case that 23 mA runs through the 2.49k R3, then there's about 60v on the 431, which is way above maximum rating of 37v. This capacitor bypass is bad news. I tried it and fried my Eros.
