See my later post where I deleted one of my previous posts ... too much on the brain and wasn't paying enough attention. I was never implying that speaker response is not impacted by amplifier impedance. I was more questioning the "constant power" paradigm, not thinking about specific implementations that incorporate aspects of voltage and current control like the ultralinear tube configuration or mixed output feedback implementations in solid-state.
As a speaker designer I can tell you that the difference between the two amplifier paradigms is significant, whether or not output transformers are involved. The relationship between the impedance curve and the speaker’s output level differs depending on the amplifier type, such that if the speaker’s impedance curve has significant peaks and dips, it will measure and sound different with the two amplifier types.
Duke