Iirc the idea is that the pink noise from the built in test tones do not use Audyssey filters.
So the point is that by running the test tone with Audyssey filters in place, the DB will be correct since Audyssey will likely have pulled down peaks, or boosted some nulls. This correction won’t be present with a “raw” test tone.
It seems very extra to me. But then again I do re-level match my subs after filtering with my minidsp. Can’t say it’s ever been that far off…