1) Yes, I believe the 100uF bypass across the Eros 6922 cathode resistor is a player, and a better capacitor is likely to make an improvement.
2) The 100uF and 2200uF cathode bypass capacitors are chosen to maintain suitable phase margins for stability in the servo bias loop. The symptom would be a very slow pumping action, which should be easily audible. But so far in my experimentation, I have never actually produced this problem. My analysis and modeling is too crude to give detailed answers (also the tube and transistor gains will vary between devices, and with temperature) so I can't say clearly what values will not work. For what it's worth, making the small one larger and the large one smaller would be going in the wrong direction.