His wording is confusing, but he does mean that the two channels, the original and the duplicate, should be panned hard left and right. The whole point of using the Haas effect is to maintain the even levels of a mono signal, but fooling the brain into locating it either left or right.
I’m not sure if you understand how the Haas effect works or not. I’ll describe it here. If you already understand it you can ignore that part. But understanding what is happening is key to knowing how to use it.
Your brain uses three different cues to determine the direction a sound is coming from: amplitude, Haas effect and frequency spectrum. Turning a pan knob does the first (i.e. sound coming from the left will normally be louder in the left ear).
Say someone is standing in front of you and hits a wood block or something. That sound will arrive at both ears simultaneously. If that person moves to one side or the other the sound will arrive at the closer ear a couple milliseconds before the other. Your brain will interpret this difference in arrival time to locate the source of the sound.
Secondly, that sound will reflect off of other objects, walls etc. causing an even longer delay. Your brain knows to ignore these reflections when locating the source of the sound. After a certain amount of delay, though, your brain will begin to interpret these as distinct echos coming from a specific direction. So the Haas effect is limited to a certain delay setting. (The article author mentions 35ms.)
The Haas effect can even override the amplitude difference. Say you’ve got cotton in one ear, muffling sound. You can still tell what direction sound is coming from even if the wrong ear hears it louder. That’s why the author of the article suggests it as a way to seperate instruments without having to pan them off center.
The third thing that affects sound locating is frequency spectrum. Normally the sound arriving at the ear toward the sound will have slightly more high frequencies than the opposite ear. Basically the opposite ear will sound a little muffled. When the author of that article talks about filtering he is meaning rolling off a little high end on the delayed copy to enhance the Haas effect.
Haas effect works fairly well in headphones, where the two ears are completely separated. In stereo speakers the effect is less pronounced, especially if you are not centered between them, but it can help to create more space in the mix. It also works best on percussive sounds and less so on things with a more even envelope. It also works better on higher frequencies and less so on lower ones. So, nice for cymbals but not so much for bass.
As an aside, frequency response actually goes even further in locating sound. This is actually the reason the pinna of your ear (the big thing that sticks out the side of your head) has the unusual shape it has. Sound reflecting off the different folds of skin will effect the frequency response of the sound differently at different angles. Your brain uses these differences to sort of fine tune the direction a sound is coming from.
These specific frequency response differences are super subtle, so they’d go beyond the ability of using it for mixing. But I once heard a demonstration of this effect in headphones. The person who made the demonstration had made a fake head with a microphone positioned where the eardrum is, then modeled all the different frequency effects. In headphones they could use this to place sounds at very specific locations around the listener.