Got it. But the original schematic says it should be 47 uF (not 0.47 uF).
Adding the cathode bypass will increase gain of the first stage. If you are going to use an outboard active preamp, you won't need the extra gain, and in that case I would definitely remove the cathode bypass cap (and the associated series resistor). I'm guessing the amp will sound better overall too, without the cathode bypass. Looking at how to do the feedback, I prefer the fourth version shown at that link (feedback taken to a junction of 1K and 330 ohms).
Concerning the amount of feedback to add, in place of the 8.2K feedback resistor, temporarily clip in a 50K pot. Hook up your bench speaker, ground the input, and then slowly dial down the pot starting at the high end. This process will slowly add more and more feedback. At some point the amp will go unstable and you'll hear thumps, whistles, chirps, etc from the speaker. That is the point of the onset of too much feedback which causes the amp to become unstable. Remove the pot from the circuit and measure it. You need a feedback resistor larger than that measured size to keep the amp stable. This exercise simply determines the maximum amount of feedback that can be applied. Now try it for the other channel, and if the output transformers have the same characteristics, you should get the same (or very close) measured reading on the pot.
As to how much feedback you should add, that depends on what sounds best to you. Try a few different resistor values (keeping them larger than the min value you determined through the above exercise) and see what floats your boat! If you play music through the amp, and adding feedback does not reduce the output volume discernibly, it means you've added very little feedback and probably not enough to do any significant good.
If you had a signal generator, you could be much more precise about it, but the above should get you into the ball park anyway.
