The specificity with a digital simulator is that the sound personality is already written in the patches/profiles (how well, depends on the simulator's quality). When you use a profile, the cab signature is already present.
At this stage, if you amplify it with a further guitar cab, it would be as if you mic'ed your real cab and sent the mic signal to another guitar cab rather than to PA.
It's not a matter of how many loudspeaker the original cab has, but of the ability to faightful reproduce that sound. That's why a good PA or a linear cab are many users' choice.
Keep on mind that onstage, you can have a wall of 4x12s but you usually put one mic in front of *one* cone... This is what arrives at audience, and what usually ends up in a record.
OTOH, you might want to kill the cab simulator in the KPA and use a guitar cab. If the resulting sound makes you happy, you're done! But all your sounds will gravitate around your real cab's signature, and won't generally be faithful to the profiles' original sounds.
How much this is an issue solely depends on user's needs and preferences. What matters most IMO is that we know what we do, and don't harm anyone