of course as the middle-class is vanishing the correspondence is perhaps not as valid as before, but everyday Americans would once have been understood as a reference to middle-class Americans.

When you have a random variable with log-normal distribution, such as income or wealth, unless the fundamental mechanism is changed to be additive (normal) instead of multiplicative (log-normal), you'll end up with the bulk of people at the low end of that curve. The existence of a large middle class was an anomaly.