Suggestions include sand in the mix to make it easy to grind, tiny bits of the querns grinding off, reed sieves that didn't get rid of the detritus--that sort of thing.
It turns out that some of their querns were granite or basalt (some were limestone) and should have been pretty durable--and I'd expect pharaoh to get the good stuff. But what was the good stuff? Maybe the abrasion was so slow that it just seemed like texture, maybe even comfort food texture. I was all set to try to figure out why they'd use such cruddy flour, but some other folk got there first.
Some archaeologists have been trying to reproduce ancient methods and figure out what ancient grains they used. There were a variety of possible strains of different hardness and variety of different coarsenesses of the flour, depending mostly on how many passes through the grinding it went through. All by hand, of course, since the circular quern wasn't invented until the fifth century BC and I don't know how long it took to reach Egypt.
Oddities abound: convex quern surfaces (transverse, that is), actually work pretty well if you use a rocking motion with the handstone to crack the grains first. Soft grains aren't quite as easy to grind since they moosh up into the grains of the stones and make it slippery. It shouldn't have been necessary to add grit to make the stones work if you use hard grains.
So. Maybe they added grit anyway. The linked article estimates 3 hours to prepare 2kg of grain. Shortcuts would have been attractive, and if you are going to sift the sand out afterwards anyway. (Or at least most of it...)
The Anasazi used sandstone tools and wore their teeth out pretty fast too.
Tooth enamel is about 2.5mm thick at the thickest. Say you wear all the way through that in 40 years, and then start wearing through the soft part quickly. That's less than 0.2 micron per day; not dramatic on a day-to-day basis, though I'd think you'd feel the grittiness. Maybe they didn't make the connection, and liked the texture? Even if someone did make the connection, proving that some other technique was better might take too long--Daniel et al's longitudinal study was only a few weeks, not 40 years.