I think most people don't realize that the cost for e-ink on a shelf versus paper is still way in the favor of paper, as you said. And that says nothing of the technical challenge that comes with adding that many devices to either an RF or IP network and ensuring that it works with existing, sometimes decades-old pricing systems. Then there's the legal challenges for displaying one price on the shelf but ringing another at the register - with paper, a human can indicate that the price label is updated, but with e-ink, there needs to be some other confirmation of change.
The cost is not in favor of paper and hasn't been for a decade or so. These devices have been in widespread use in Asia for a long time. Best Buy went to them a few years ago, Walmart is moving to them.
The devices are often optical, with overhead IR transmitters. The fact that you don't know that shows how little you know about the market and thus should not be talking about it.
RF is also used, yes - various protocols ranging from proprietary to bluetooth LE.
Even if only one price per second can be transmitted, that's 3600 items per hour, which means that in 24 hours, 86,400 items can be re-priced. You'd need several employees working non-stop to even hope to hit that rate.
As for the legal challenges they're able to point to logs as well as information on the digital display showing when it was last updated. Also, updates are not necessarily live - they can push out updated pricing info and then activate it at a specific time or via secondary command.
In some places, the calculation is clearly going in favor of E-Ink tags. I assume the lower management effort you mentioned will be a part of that.