Modal behavior always introduce cognitive overhead. It's something that you have to track, and it introduces the possibility of getting it wrong. To your point

Modal behavior always introduce cognitive overhead. It's something that you have to track, and it introduces the possibility of getting it wrong. To your point, that overhead can be reduced through various UI techniques, but it cannot be eliminated. It's an extra bit of information you need to track. It's an extra check you need to make before reading the situation (and critically, it cannot be done in parallel). Switching modes is an extra step you need to take before performing an action. These are all additional points of failure. In a high-stress, high-stakes situation that's something you want to avoid. That's not to say that modal UI's are always bad and never should be used. I'm saying that there's a tradeoff, and that tradeoff is not worthwhile in safety-critical applications.