The reason for 2.83V is that you can apply 2.83V to a loudspeaker.
And the actual impedance may not be a standard 2,4,8 ohms.
By applying a constant voltage to all models, it makes it easier to compare actual sensitivity with a constant source.
As with other things audio, a simple answer results in a wrong answer.
Yes, it is simply easier to test everything with one voltage regardless of the nominal impedance you designate your speakers that may or not be 2,4,6, or 8 ohm.
It's been about a dozen (or so) years since DSL changed sensitivity ratings from 1W/1M to 28.3/2.83v for everything regardless of impedance, though still recall the discussions regarding the "1W/1M" SH50 with a 6 ohm nominal impedance and a 3.5 ohm minimum using 28.3 volts at 10 meters, and a dozen crossovers for John A. Chiara (aka. Blind Johnny).
Difficult to relate to sensitivity ratings when the "watts" used for DSL's "power handling" still are the same as the drivers installed are rated at using voltage specific to their nominal impedance.
Just math though, everybody knows that you simply subtract 0, 3, or 6 dB from the DSL specs to convert back to the olden days when we thought in terms of watts rather than volts. I remember those 40volt amps were really heavy ;^)
Art