That makes me realize, it would actually be an incredibly fun job to be in the team that manages the simulation/emulation stack that people tinker with to learn about this thing.

Going on a lot of presumption here (and a bit of optimism...), I would presume "don't brick it" is backed up by an appropriate level of funding to cover the training people receive before they get to touch production. In an optimal scenario (don't be wrong... don't be wrong...) this would cover component-level electrically-accurate (because radiation) logic simulation that everything gets tested on first.

NASA doesn't mess around with simulators. I've read accounts from astronauts and ground controllers where they say their actual missions feel anticlimactic after all the insane things they had to do in the simulators. Gene Kranz managed to crash a simulated Apollo mission into a mountain.