why is PSU regulation arguably unnecessary?
Usually a power amp has power supply rejection, it generates its output with reference to its input within the capacity of its power rails but without reference to them. So the ripple implicit in a transformer/rectifier/capactior setup doesn't appear on the output (much...).
A conventional regulator for this arrangement is much like a power amp, you give it a constant input reference voltage and it generates and holds the desired output voltage without reference to the raw supply (pretty much).
The reason power amps rarely use regulated power supplies is because of the currents and voltages involved. The regulator is about as expensive as as the power amp itself, requires as much design input but from a much smaller pool of experience, and tends towards wild oscillations when driving anything other than a resistor...
A power amp with a regulated supply is like two power amps in series, the raw supply passes through a device controlled by an error signal, then through another device controlled by another error signal, and then to the loudspeaker. It might be worth considering combining the error signals, and spending more effort on the power amp itself rather than on the supply.
OTOH there are some very fine amps with regulated PSUs, engineering anything is all about compromises rather than abstract purism.
I think one of the best arguments for regulated PSUs is to implement non-intrusive current limiting and DC offset protection. You can shut the PSU down in the event of a short circuit rather than have relays or other interference in the signal path. This is the kind of functionality that DIY amps can live without, but real customers probably get quite annoyed when their new amp goes up in smoke during installation because they were careless with the speaker wires. And then because they didn't quite believe it they blow their speakers up too....
Paul
