How do you control and assemble, not to mention cool and extract energy from as well as, if you believe Rossi, take heat and electricity away, oh and... it does require a DC supply (why?) And there are 50,000 tiny devices CONTAINING A PLASMA to make a megawatt plant?!
I do not think the device exists, so this is a hypothetical question. However, I do not see why it would more difficult than controlling thousands of fuel rods and control rods in a fission reactor, or thousands of small battery cells in an electric car. A robotic plant could manufacture thousands of small plasma devices in a single permanent array, with integrated control electronics.
Vacuum tubes are plasma devices. Toward the end of the vacuum tube computer era, there were reliable machines with tens of thousands of tubes. If the technology had been further developed, with small tubes integrated into a single memory or logic device, that device would be highly reliable. We know that because, in fact, that's what happened. Charles Spindt of SRI and others developed chips with thousands of microscopic integrated vacuum tubes, in microscopic holes, for specialized high speed signal processing. For a while, flat panel displays used similar technology.
As I said, if we pretend these Rossi devices actually exist then . . . Assuming the plasma devices has significant advantages over the macroscopic single reactor e-cat gadgets, the customers would migrate to the small plasma one over time, just as Tesla has migrated to automobile batteries with many cells. Your feeling that macroscopic devices are inherently easier, cheaper or more reliable is based on 19th can 20th century manufacturing technology.