I've done some testing with this and the example that I did the biggest test with was 6mm 109 ELD-Ms. I weight sorted through 1200 of them to get 30x each of the "high" and "low" brackets. Low was 108.7gr, high was 109.2gr.
The high and low MV separation on 30x ea. was less than 1fps on the average velocity. Either 0.3fps or 0.7fps I can't remember which. There must be some self-correcting physics going on with the powder burn because if you do it proportionally scaled to the weight difference it should have been more like 19-20fps.
MPOI and dispersion over that 0.5gr spread was undetectable.
The only place I could see it mattering, and it was very small net effect at 1000yd was with drag. Mass is a direct 1:1 scalar on total drag. If you have bullets that are 5% different in mass, but otherwise externally/physically the same, you will have very near 5% difference in drag. This will hardly manifest to anything inside of 600yd, will start to show beyond, and the effect just grows the further you shoot. For example, in my situation it was a 0.4% extreme spread in mass and drag on the averages between those two 30x samples. Remember that those two samples were picked over 1200 rounds of sorting and the vast majority of bullets would be much smaller spread.
The weight distribution in most cases is a normal distribution so you can figure what the SD math will play out as for what you can expect to see in X number of rounds. Add in inherent drag variability, MV spreads, mirage conditions, etc... I'd say it's not worth your time in PRS. On paper for F class or BR scores and how close those can get, it's probably worth doing. YMMV.