Differences in delay for different frequencies are non-existing for amplifiers. They amplify the time domain signal, containing all frequencies therein. They do NOT split the time domain into different frequencies and then amplify each frequency seperately. If a complex note, containing different frequencies, starts at a certain moment in time, it will still start at a certain moment in the time domain after amplification - it will not be 'smeared' in the time domain due to frequency dependent delays.Ok, it turns out what can be read overall: phase deviation is something pretty complicated and therefore many people don't understand it and therefore don't want to care.
Each device needs some time to process a signal (like a sine e.g.). If you have an ideal linear system, all frequencies need the same time to be processed and therefore are leaving the device at the same time +x. No problem there.
But: an amplifier e.g. isn't an ideal linear system as it is de facto a low pass filter. Therefore, the time for passing the device differs for different frequencies. Those differences are vital for the sound stage e.g.
Therefore: what I'm missing is a graph delay vs frequency e.g.
Here you can get the background:
All about sound and hearing (german)
Some theory (german)
Some more theory
And for those who still negate the problem: here you can hear it with your own ears:
Practical with examples
What does this mean if one amplifier is compared to another one: they might sound equal if they have same usually measured values - but they might sound different, if they differ in group delay, which usually is not measured.
That's why I want to see for each device the information about the group delay.
A frequency dependent phase shift may occur, but a phase shift is NOT the same as a time shift. A phase shift causes specific frequencies in the onset of a note to start with a different phase, they will not start at a different time. Frequency-dependent time shifts are audible, frequency dependent phase shifts are not. (A time shift does produce a phase shift as side effect, which causes the two to be confused.)
The audio examples in your link use (pretty big) differences in delay time between frequencies, which is completely unrealistic for an audio amplifier.