back away from the distinction again and again. It's not programming vs not-programming, and it's not about building stuff in meatspace vs. building stuff in memory.
It's about public vs. private effects and risks. People who build stuff for backyards = not engineers. People who build stuff for the public square = engineers. People who build a bridge over their own stream themselves = not engineers. People who build a bridge over a public byway for heavy public use = engineers.
The same ought to be true in codingland. If you are building critical public infrastructure or software that many people must use in order to participate in society, then you should be licensed and bonded and held to higher standards. If you are building software that is for personal, private use, then this is not the case.
Note that I am talking about instance of execution, not instance of code. For example, MS Windows would not be "engineering" code because each instance of execution is private, while the code inside, say, a traffic lights system, aircraft control system, or automobile is "engineering" code because these instances of execution happen in public, with public effects. (If MS Windows was to be used to run, say, a battleship, then it ought to be a separate "engineered" version held to higher standards of scrutiny, professionalism, and liability.)
We don't certify and license people as engineers because they might touch wood and steel we think wood and steel are special or important, we certify and license people as engineers whose single instances of wood and steel construction will each touch (and possibly put at risk) many lives.