A long time ago, as a very low hours PPL holder, when I did all my ME, night and instrument training on a PA39 Twin Comm, most of the time, single engine work was done with simulated zero thrust, but my instructor was very specific about my doing one in flight shut down, a good period of time genuinely on one, and then, the most thought provoking, the restart. We didn't lose any height, but I found my hands had to be flying around all over the cockpit to keep things stable, going in the direction it was meant to be while doing all the other things required for the restart. It was a bit like demeted wall paper hanging, with changes of hands on the yoke, and such like, just to get it started, warmed up safely and then back in balanced flight, but it demonstrated only too clearly that there was a lot of work involved in what in theory sounded very simple to do.
The end of it all, and I was happy that we spent a lot longer than the mandated 5 hours to "do" the ME rating, was that I spent a couple of years doing a lot of hours all over Europe in what was recognised as a very slippery hot ship twin, most of the time single crewed, quite often IFR and IMC. and I thoroughly enjoyed it most of the time, and the few times I didn't it wasn't the fault of the airplane.
What was even more interesting was several years later going over to the States to do a CPL IR ME. I got put with an instructor who probably had about 20 hours total ME time, and at that stage, I had close on 300. It didn't work, the instructor was not comfortable with the airplane, and even less comfortable with the manner in which I was totally comfortable doing things with it that he wasn't. Cue change of instructor, which solved the problem very rapidly.
Making the student demonstrate handling to VMCA is risky, in that the student's responses are likely to be a lot slower than the response needed to keep things in order, especially on the hot ships like the 310 or the PA39, both of which have a "reputation". We certainly did work close to VMCA with simulated zero thrust when doing some of the IR work, to make sure that I was capable of handling the aircraft at the extremes of the envelope, in the same way as I was made to do work close to the VNE end of the speed limitations, to make sure that I knew what to do if I got there by mistake, which it wasn't hard to do on the PA39. It was all about making sure I could recognise the direction of trends, and take positive timely action to change the trend before it went out of limits, which was (in my book) good instructing.
So, I've flown very close to, but not actually at VMCA, and done the same with VNE, for the same reasons, and as far as I'm concerned, that was and is the safest way to deal with it. Anything beyond that should (if possible, and unfortunately on most GA types it's not) be done in a good quality simulator, and by that, I don't mean Microsoft Flight Sim xxxx, as that's not got the necessary fidelity in flight modelling to allow for extreme of the envelope work.