Dark matter is believed to somehow cause friction based on some things I've read (don't have link yet), and it hangs out at the outskirts of galaxies,
Errr, no.
DM is distributed, probably spherically, around galaxies, with the highest concentration in the middle of the galaxy. It's not as closely concentrated as baryonic matter (because it's got fewer interactions, that's what the "dark" means), but when people model the mass distribution of a gravitational lens, for example, the put the centre-of-mass of the baryonic matter (visible matter) and the dark matter at the same point, and the spread function for the distribution of dark matter around that point a factor of (several ; it varies ; often it's explicitly a variable they try to optimise for) times that of the baryonic (visible) matter. It seems to be an accepted procedure.
The alternative take : what force is it, which DM responds to but BM (baryonic matter, visible matter) doesn't, that constrains the DM to the outskirts of the galaxy?
The reason that the outer parts of Joe Random Galaxy rotates faster in a DM model than in a BM-only model is that the DM+BM model has more matter at a large radius from the (mutual, DM & BM) centre-of-mass model would have. Both are similarly distributed around the c-o-m, but the BM is more concentrated to the core than the DM. So, unlike a Newtonian model of orbits in the galaxy, with essentially all the mass near the centre (e.g. the Sun (98.5%) in the central million km of the Solar system ; Jupiter (1% of the Sun) at 750 million km, and the remaining 0.5% smeared between 1 million km and about 1000 million km) there appears to be mass out at (say) 2000 million km. But really, you need to integrate the mass distributions against radius in the two distributions.
Yeah, I had to read various papers about 5 times to work that out. It's not DM in an annular structure around the BM part of the galaxy, it's two overlapping distributions with a different central concentration.
where early passes typically occur. Thus, I'm referring to what might be called "non-baryonic friction".
The term is "dynamical friction", and it's a property of granular distributions of non-light particles under gravity. It's nothing unique to DM, or BM. Again, it was a several-paper comprehension dawning - can I recommend again Mike "PlutoKiller" Brown's Caltech course on the structure of the Solar system, which is available every few months on Coursera. The course is around 20 hours of lectures, and you'll need another 100 or so hours of reading course material to do the homework. But it is well worth the investment.
PlutoKiller's use of the concept is in analysing the motion of "planetesimals" (say, 1km blobs of dust and ices not zero-mass points) accreting onto larger planetesimals, which become planets. Or, as the surface density decreases, stay as independent planetesimals et voilá - Kuiper-Edgeworth Belt. Dynamical friction keeps the orbit of the planetesimals circular while they're still accreting, but once the distributed mass has accreted into a few (dozen~hundred) point masses, they can start evolving in both eccentricity and inclination, which leads to a period of hierarchical growth, and Bang! - Moon-forming (Venus-reversing ; Mercury de-mantling ; Mars-Polar basin, Uranus side-flipping) impact as a general outcome.
Really, that course is worth the effort. It will make a lot of things clearer.
Oh, this paper defines "merger" as galaxy positions within 20kpc of each other, then goes on to say the results are not particularly sensitive to them varying this parameter. 20 kpc ~=70 kly, which is not that much further out than the Solar System's position in the MW. It's a pretty central hit. Body-shot, not limb-shot.