Aside from ROME rumors, there have been leaks on the Ryzen 3000 series (Zen 2), I read somewhere these mainstream CPU's will be clocked closer to 4GHz? I would assume that server technology trickles down to consumer CPU's and there would be higher clocked server CPU's. Or maybe I'm missing something.
I think you have that backwards, and the simple explanation is that the servers are denser and have more cores. Consumer stuff isn't as tightly constrained (for instance Threadripper gen 1 had half the cores for basically the same TDP as EPYC, and the consumer stuff had 1/4 the cores but a 1/2-1/3 TDP. So they can clock the consumer stuff higher). It might would be possible to clock the server parts the same as consumer stuff, but you'd need adequate power delivery (power supply, and motherboard capable of it), as well as adequate cooling. To hit the 4.3GHz range, the 8 core consumer chips needed 105TDP, and I believe took a bit more than that in reality (oh and I don't think that was all 8 cores at that speed either). So consider what it would take to 32 cores to that level. You'd be pushing 500W. But EPYC has what a 180 or 250 W TDP. But also clock speeds and power is on a logarithmic scale (so you can see big power decrease by relatively smaller clock speed decreases).
With Rome it appears that EPYC might have 8 times the number of cores as the consumer chips (although people expect it'll actually continue about the same as Zen 1, where EPYC had about double the cores of Threadripper and 4x the number as consumer Ryzen, so we'll get 16 core Ryzens at some point, but for now they've just touted the 8 core chips).
A fair amount of expectations are that consumer Ryzen will hit closer to 5GHz. I think people expect EPYC to be closer to 3GHz.
The reality is that modern CPUs have a lot of load leveling so it kinda depends on a variety of factors, especially cooling, for what you can achieve. The CPUs can basically go "I've got this many cores being used, this is the TDP, these are the temps" and then adjust clock speeds from there, and as temps go up it'll adjust clock speeds. So you can get short boosts up but sustained clocks would likely be lower. And all cores going full tilt will lower the clocks.
7nm is letting them increase the density and also push clock speeds some, which is why its increasing over Zen 1. It basically gives them 2x the density for the same size chip (hence why we're getting roughly twice the number of cores with Zen 2), and about 1.25x performance for the same power (for the 4GHz consumer Ryzen that would put it around 5GHz, although its not quite that simple as the individual cores are being expanded some, and not everything will get the full density boost; and in AMD's design, a good chunk of things is in the I/O module that isn't being produced on 7nm).