I've heard that one of the main reasons for using a staining developer is that the stain "mask" or "veils" the grain, reducing apparent grain in highlights. I've also heard that the stain is "grainless," which would seem to be the reason it masks the grain. Apparently, old-timers used to bleach the silver out of pyro negs to get a grainless image.
I tried to put this to the test by making a strong hydroquinone-based staining developer, overdeveloping the hell out of a piece of Delta 400, then bleaching out the silver with ferricyanide. The result was quite disappointing: the remaining stain was very grainy, and not sharp either.
It would seem that the grain-mitigating effect of staining developers is something other than a "grainless" stain. Maybe the "grains" of the stain occur between the silver grains.
It's also possible that the test has no relevance to traditional staining developers that use pyro and catechol.
The developer was:
10 ml phenidone 1% in alcohol
20g sodium carbonate monohydrate
15min at 75F