Thanks for that thorough explanation. I figured it was in the display process and I have also seen the flat line condition you described. Now I know what is causing it.
Although the example compression of T/Div may be inappropriate while looking at a square wave, it could be appropriate while looking at the modulation of an AM modulated signal. Now I understand that while looking at an AM modulated waveform, you could enter this region of inaccurate displayed waveform amplitude and the amplitude should be confirmed. A quick range adjustment of T/Div steps to confirm that the amplitude remains constant, would be appropriate.