> A single N-type active device with a single-supply current source (limiter) load, from the positive rail, will perforce have constant power supply loading.
Well, yeah, but... who does that? (Except a few hi-fi nuts.)
A one-6V6 gitar amp will have a choke (transformer) load. Over a single audio cycle, current is conserved. But over the duration of a gitar note, current can change about as much as it wants to.
In very-linear use, power output far below clipping, no stupid/clever tricks, the 6V6/choke current will be constant.
Just below clipping (large current swing), and without feedback (gitar amps often minimize or omit NFB) the current will rise due to plate rectification. The 6V6 is not dead linear, the up swings are bigger than the down swings.
Just above clipping, THD is still not over 5%-10% so this may be the "book condition". However the grid is peeking over grid-current threshold, rectifying, and (assuming R-C input coupling) knocking itself negative. Plate current falls. This could be where a one-6V6 shows "sustain" (it is acting as a not-very-very-dirty Limiter) so it may be routine gitar tone.
Up to "gross" distortion, the current variations are like 5%-10%, so supply "sag" resistance has only a small effect. IMHO the glass rectifier fad is more about push-pull amps which have 20%-100% current variation from small to large output, not the one-6V6 and kin.
With adequate NFB, current is (nearly) constant up to clipping, not even those 5%-10% shifts.
Far into gross distortion, current is dropping because we have exceeded the tube's current passing ability. Output level will not rise just because the plate voltage rises. It may rise because G2 voltage rises which increases the tube's current limit. But it would have to rise a lot to have any audible effect. Such effect is probably opposite to what we want (though who knows?). And since we often need R-C decoupling of G2 to control buzz, there is a time constant that may fight the music.
I think hollow rectifiers are an obsolete abomination. I think the last time I left one in an amp, it was a series-string radio converted to an amp and I needed the rectifier drop.... no, in fact I remember wiring-up a mess of 2W 47Ω resistors to bypass the rectifier heater and feed the 50C5 and 12AV5 heaters. I hate hollow rectifiers. I think they are bad. They are an utter waste of heat. I know they fail more than most tubes. I think rectifier design was a hasty afterthought: most tube rectifiers are pretty bad performers compared to other late-1950s tubes. I think any "good" they can do is mostly imaginary, and anyway you can probably get the same effect some better way.