The major problem isn't the peak, is the big distortion here, you may not want the peak, would be nicer to have it at lower freq, then you can damp it. The resonance is there because the inductance has changed, the ~50% THD rising fastest around the resonance freq is because the source reactance (capacitive) at that point is the same as the transformer reactance (inductive) and as the transformer impedance is changing with level vs a comparable impedance feeding it the high distortion is there. I almost certain now that using a bigger cap there would end in less distortion, how big and what effects will it have in the freq response is another thing to look for. I wouldn't want that the freq response of my design is being limited in the last point of the chain, anyway we are needing a bigger cap here, we will have higher resonance, so it will need more damping, now it has almost no damping, but the output impedance of the tube stage which is rather low. How low, I think I would go with twice the capacitance is already there, I think it might end up close to -50dB THD @30Hz, +26dBu (based on the plot which has that figure for 60Hz now) and probably about 40dB for 20Hz, I think I could live with 1% THD in here at max output level.
JS