Where does it say that anywhere in Reynolds' paper? He didn't call it the Reynolds number, so I know for a fact that he didn't use "Re" in the paper anywhere. I have his paper in front of me and I don't see anywhere mention of specific values of the Reynolds number (which he refers to only as [itex]\frac{\rho c U}{\mu}[/itex]). Were you referring to a line in another paper that cites Reynolds original paper or am I just missing something?
Meanwhile, any fluid mechanics textbook covering pipe flow will tell you that pipe flows are usually doomed to become turbulent starting in the range [itex]2300\leq\mathrm{Re}_D\leq4000[/itex]. Of course laminar flow is maintained for some distance downstream because the pipe still has to go through the transition process, but the general rule is that depending on the pipe roughness, that is the range of [itex]\mathrm{Re}_D[/itex] you expect before the flow will eventually transition. On the low end of that range the pipe will transition quite far downstream.
This actually isn't in contradiction to what you just said about having laminar flow present as high as [itex]\mathrm{Re}_D = 2 \times 10^4[/itex] because all your quote says is that laminar flow is maintained at least up to 100 pipe diameters, implying that it does still eventually transition as predicted by the common rule of thumb. Pipe flow is perhaps the only flow that follows such a simple rule of thumb like this that we know of.