I'll just mention a couple further technical issues.
Using a cathode resistor to set bias is sometimes called automatic bias because it provides negative feedback at DC to stabilize the plate voltage. Normally that would be the reason to avoid fixed bias such as an LED. But with a current source plate load e.g. Speedball in Crack the cathode current is constant so a resistor cathode bias will not supply the bias stability that it does in a normal resistor-loaded triode amplifier stage, so there is no reason to avoid it.
If the resistor is unbypassed, it can sound comparable to an LED. If the plate load is a current source, and the following load is high impedance as well, this won't affect the gain. But the negative feedback it provides in the audio range causes the plate or source impedance to increase. As long as that higher impedance is not a problem, there is no reason to avoid the resistor. In the new Stereomour II, I have increased the plate current in order to reduce the resistor value to where it does not make the 12AT7 plate resistance too high. I did this to get a bias voltage between the HLMP's 1.55v and the TL431's 2.50v. Engineering is always a matter of tradeoffs; the meaning of the word "best" varies with the context.