gm is going to be ~10 mA/V; R_load is 10k or gain of 100. Divide by two for the differential circuit, and we get V/V of 50 per phase. Looks like a little more than a volt to deliver the 60V or so to the 2A3 grid. We could try a lower load and take a bit away, but remember to raise the g2 the correct amount to keep a nice clean area to work the load through. The 12BY7 is a nice tube, and the required data to make these g2 voltage determinations is available. I am tempted to try its DH cousin...:)
cheers,
Douglas

remember also that the grid choke is going to offer an eliptical load at the LF extreme. Lowering the load ( to lower numeric ) would alleviate that sort of behaviour a bit.