I'd speculate that these Master Programmers would have to be sick puppies if they were doing this for sport. "sport" like the Roman Coliseum, maybe. Or maybe they're building Purgatory-2.0. A lot depends on the goal of the simulation.

But that's a bit of anthropocentrism, assuming that the simulation is local and all about us. That would be a Matrix-style simulation--on-demand, as needed, when looked-at--which is actually a highly rational way of running a simulation. It not only uses the minimum required resources, but it includes only the minimum data to answer your question. If you want to know whether Nero will take the red or blue pill, you don't need to simulate all 10528 or whatever hydrogen atoms in the universe, you just need to simulate the subset in Nero's vicinity. To be manageable, a simulation has to limit itself to relevant data.

I doubt the utility of running a whole-universe simulation like you posit, to answer general physics questions like the effect of varying Planck's constant. You can do that with a tiny subset of the whole universe, because after you've made your first few trillion observations, there won't be a lot more to learn. If you've seen 10128 protons, you've seen 'em all.

And even if they knew we were here, they could still just as easily write us out of the program, just as a novelist could eliminate a minor character because he does nothing to advance the plot.

Then I guess we'd better get our shit together and start advancing the plot, eh?

But don't worry, if it happens, reboot or shutdown will be painless.

I promise.