Real hardware sounds great... but it would quickly become unwieldy. Other than the "tangible" aspect of real hardware... what's the benefit? Creating the machine in HDL is really the "same" as hacking it together physically... but without the mess.
Keep in mind that hardware is only 1/2 of the class/book.
The other problem I see with a physical hardware implementation is I/O. It wouldn't be very hard to do the CPU and RAM, but the ROM, Screen and Keyboard are problematic.
What's the tool path to get code into the ROM? Screen will need to be dual port RAM with some fairly sophisticated circuitry generating a video signal that can be displayed on a standard monitor. Keyboard's even worse.
The brilliant thing about this course is that the authors came up with a very simple design that can be understood by people just starting out on the computer program/engineering path. Using the simulators keeps everything in the theoretical domain so there are no real world gotchas to distract from the direct sequence to the hack computer.
(When the student proudly shows us his two term and-or mux implementation, let's tell him that it's bad because it has a hazard. Then explain how adding another and gate fixes the problem even though it doesn't change the truth table. Didn't we just teach him about Boolean algebra and how it let's us prove that the two circuits are equivalent?)