My personal struggle with functional programming is how difficult I find it to be iterative. I've very frequently backed myself into a corner and ended up having to rewrite everything, whereas in C, even when I miss badly on the architecture, I'm still usually able to reuse some big chunks of what I've already done in the end. Maybe it's just inexperience.
Deep learning with backpropagation? That's the chain rule of calculus applied recursively over a network, as described by Rumelhart, Hinton, and Williams. We owe LLMs and emerging AI to recursive ideas.
The internet? It's a recursive structure made up of an arbitrary number of interconnected servers. We owe the information era to recursion.
Database queries and programming languages? We prove correctness by reasoning recursively. We architect entire systems of arbitrary size, with correct behavior, simply by specifying the properties of their element types.
Sure, you could probably block recursion within the scope of a program's compilation. But when you're developing scalable systems, recursion is their natural language. The powerful recursive algorithms for these systems are symptoms of their scalability, not mere implementation choices. When forced, stacks and other recursion workarounds are obstacles that get in the way of innovation.
Plus, real-world systems aren't single programs compiled together. When microservice A calls microservice B, there's potential for recursion, regardless of whether the programming language of either microservice supports it. Programmers still have to think recursively. No programming language annotation is going to change that.
The Great Pyramid contains a combinatorial explosion of smaller pyramids inside it. They made it out of blocks. Forty centuries look down upon us.
It's way beyond brain-teaser for me, it's the sharpest tool I know of.
It's not clear if you mean "the functions that you write" or "the functions that other people (library providers? Colleagues? Juniors?) write.
The functions you write do what you tell them to do. If you don't tell them to recurse, then they don't. If you do tell them to recurse, then that has whatever frame optimization or boundedness guarantees that you bake into it (which is the same responsibilities you take on writing a loop just packaged in a slightly different form).
The functions black-boxed by a library aren't something you have a lot of control over, nor in most cases are those black boxed by colleagues. Your concerns sound as though they target these cases, though your preferences really have no traction here.