There's been metabolic studies that show that this isn't true. Comparisons of total caloric usage of completely sedentary people and people who have high exercise load are indistinguishable. There is a large difference among individuals, but not correlated to exercise levels. Sedentary people who start training hard will have a spike in caloric usage for a few months, but their body adapts and calorie burn returns to the same level that it was when they were sedentary. This was new research, so there wasn't an explanation for it. The authors hypothesized that it could be that the body reduces caloric spend on other things, like stress responses, when it is adapted to high exercise levels/ They did note that some extremely elite athletes can temporarily increase their caloric burn (think Michael Phelps eating 10k calories per at some points when training for the Olympics) but its not something most people can achieve or sustain.
If software engineer productivity basically doubled as is being claimed in this thread, I think you'd see companies scrambling to lay off everyone else in an effort to hire even more software engineers. They'd be by far the most valuable and productive employees at every tech company and you'd be foolish not to have as many as you can. I'm being a bit facetious but throughout history when a resource or profession takes a dramatic leap in efficiency, the demand for that thing rather than decreasing as is predicted here, only increases since it has become far more valuable & effective.
My company's CEO comes from the sales world, and I imagine that's the case in many companies making these RTO decisions. His idea of getting work done is getting everyone in a room together, having some handshakes, sitting down, and talking something out. This is not what getting work done looks like to software engineers, and many other IC positions. The blanket RTO policies come from a lack of understanding how other people & roles work best.