The Mullard 5-20( I think ), with its input pentode direct coupled to the LTP arranged twin triode is nice, but with its ~100V loss to the cathode load, it isn't practical.

Swap the cathode voltage allowance and resistor for a CCS and -10V negative supply opens up a lot more headroom. Swap pentodes( which do need a carefully matched g2 voltage and load choice ), and there's headroom to spare for even low-mu tubes like the 2A3.

The input stage being a pentode offers a lower capacitive load for the linestage. It is also a more static capacitance as Miller is not making an appearance due to the effects of the screen grid.

It is ridiculous( IMO ) to suggest that a *REALLY GOOD* amp can be done from a circuit compromised for simplicity. Some circuits are quite simple. I would invoke Einstein's rule, as simple as is required, but no simpler...or something like that. Anyway, it is not so difficult once the holes are drilled and the Iron is mounted.
cheers,
Douglas