I could see the Cw in parallel, or adding itself to the driving stage's capacitive load( grid-cathode, and grid-plate multiplied by gain, or Miller effect ). The units just don't add up. Capacitance times capacitance does not yeild capacitance but what I suspect would be a meaningless quantity. Mass*length^2/time^3 is a means of expressing power...I'll leave the capacitance exercise to the student...:)

The choke's Cw would present a load to what ever is driving it. It's not referenced to the grid, but elsewhere in the circuit, so I have additional doubts on the 'multiplied by Miller' that you attribute to Gordon. I think a schematic with all the pieces represented, and the math behind them would be of great use.

The cascode and pentode offer a high output Z device for use in the E-Linear circuit. The circuit has a fair amount of short path NFB, and the additional gain is useful there up to a point. Considering how easy it is to build a linestage with gain, the amp's gain is pretty much a non-issue within reason.

On the grid current, I have some similar suspicions about chokes. Signal caused grid current is AC in nature, and since the choke only pays attention to AC...one should be very careful about using too much.
cheers,
Douglas