Are they? The goal isn't to draw a hard boundary it's to create layered defenses which increase the difficulty and reduce opportunity.
If instead of open access you need to tailgate on a limited set of employees, that increases difficulty considerably and makes the opportunity much less common.
Real security analysis works this way: you don't assume you can build a wall which is never breached.
I'm not sure the problem with card readers is that it's security theater, so much as it's a really inconvenient place to put the security check so the added security is not at all worthwhile. Putting the card readers on each floor so that you need to swipe immediately after exiting the elevator separates the security check from all the ordinary functions of an elevator, and splits up that security checking across a large number of floors instead of a smaller number of elevators. Everyone still needs to go through the same number of layers of security checks, but those checks aren't being imposed directly on top of what's already a bottleneck.
So the lesson that should be familiar to software developers is: retrofitting security (in the elevator) instead of planning for it from the start (with space for access controls between the elevator and the offices) will tend to make the security more painful than it could have been.
Deviant Ollam talked about elevators a lot. You can have the fire marshal's key or an elevator technician's key and know where the secret panel is to override the controls on any elevator, take it to any floor you want, or even between floors, or tag it out of order and hide in there until everyone else goes home. Any elevator shaft should be treated like a door to the outside world, you never know who could come in from the elevator, even if there's a key card.
You're missing the part about layered controls which is what elevator access control is.
Limiting the number of people who can incidentally travel to and loiter outside your primary office entrance is one level of control - it raises the visiibility of interlopers, and creates an initial barrier to even being near the objective.
Everything else you've listed though is also just a big list of additional breaches and crimes someone would have to commit before they can even get near to the target in the first place - i.e. they're relevant, but if you're ingress plan is "steal a bunch of elevator access keys" then you've already excluded a whole host of simple crimes of opportunity.
If instead of open access you need to tailgate on a limited set of employees, that increases difficulty considerably and makes the opportunity much less common.
Real security analysis works this way: you don't assume you can build a wall which is never breached.