It's all about averaging.

A static tone will dwell on the peaks and dips produced by natural room modes, and the resulting measurements will read exaggerated peaks and dips in response while pink noise or warble tones do not stay on one particular frequency long enough to directly excite these modes. In other words, pink noise and warble tones average out the effects of these irregularities in response, and the resulting curve matches how the ear perceives these irregularities in response better than when discrete tones are used.

Measurements made in anachoic chambers that are used at speaker manufacturers (such as at Altec and JBL), do use discrete tones and sweeps but the difference is that an anachoic chamber does not reflect back sound from it's walls - it is essentially the same thing as doing measurements outdoors. They therefore are plotting the response of the speaker only, not the effects of the room.

This ear/measurement correlation is why we use logrithmic scales when referring to sound pressure and other audio measurements instead of linear scales. If we used a linear scale, the measured curves and how those curves are actually heard by the ear wouldn't match. A linear speaker frequency response curve for instance would look far worse than the ear would hear it.