#1. Will it work?

Yes
 
 #2. Will it pop like a son-of-a-gun if switched on the fly?

Yes, anytime you switch DC voltages in an AC circuit there will be some kind 
of  pop.
Maybe if you used a small resistor across the filter cap, at the end of this, 
to keep an extra current load on the filter cap,  that would allow the 
voltage to ramp up for less noise.
 
 #3. Do you think that the added cap should be 10uF or 20uF (in the
 high-voltage mode the caps would be added to the B+3 cap for a net value
 of 30uF vs. 40uF)?
 
It's already filtered so 8-10uF/450v should be fine.

 #4. Is there a better way to accomplish this?

Probably, but I can't think of a good one and I don't know if it would matter 
that much.
I wonder if it would be better to use a parallel resistor across a seperate 
PI B+ feeder and switch it in and out to change the voltage drop across it, 
instead of throwing the whole resistor in and out? 
If you wanted to loose an additional 100v to the tube, while it's pulling 
about 1.5ma,  then you'd have to add 68K of resistance to the circuit.
If you used two parallel resistors with one switchable, then when they are in 
parallel the resistance would have to be 68K be lower and there would be much 
less voltage drop... 100v less drop. Kinda the same thing backwards.
I don't know if that would be less noisey or not and maybe just a large .1uF 
cap across the switching resistor would absorb some of the snap.
Just a thought.... 

Bruce

Mission Amps
5563 B Gray St.
Arvada, CO. 80002
1-303-423-4103