Biggest impact should be on measured clean power output, and how high you have to turn the volume control to get to that power output.
... I understand and acknowledge your rationale (that reducing the bias voltage acting to limit signal swing at the grid and hence reduce power output), but it doesn’t seem to be working like that ... the peak voltage being negligibly different to that achieved at high idle current.
If I've understood correctly, this sounds like my original post should have been, "Biggest impact should be how high you have to turn the volume control to get to the same power output."
Which makes sense as peak plate current is set by the supply voltage, OT primary impedance & tube characteristics.