Indeed, in those cases one just does not care about performance.
Yet there are cases where one does. Imagine an orchestration system written in Java -- with lots of threads (perhaps because it might be a thread-per-client affair, or maybe just NCPU threads), with a large heap (because Java), and launching lots of small tasks as external programs. Maybe those tasks are ssh commands (ok, sure, today you could use an SSH library in Java) or build jobs (maybe your app is a CI/CD orchestrator). Now launching external jobs is the core of what this does, and now the cost of fork() bites.
Let's just say I want to design a solution involving spawning a buttload of procesesses and pass messages back and forward. Roughly when does fork efficiency become something other than of academic concern? 10 processes per second, 1000, 100000? What does the inefficiency look like? Nothing? A stutter you might not notice? Through to everything grinds to a halt and you can't login to the box and neither will the oom killer help you.
Until you quantify on some level, even very roughly, what the observed issue is, when you see it and how it degrades, that you're trying to optimize it's just urinating into the breeze. We might get lucky is the best outcome. The chances of it being a really good outcome are pretty limited. decrying something as "inefficient" based on big O or whatever is just meaningless until we actually do it. [1]
[1] selection sort is O(n^2) and can totally dominate O(n log n) algorithms in actual time and cycles spent depending on circumstance. We have to specify, it's not something that can be shortcut because it will likely get a terrible result.
The first link didn't even have units on its numbers(!) I assume they're milliseconds. When does that scale become something one would care about at all? Not launching a gui process. Not a shell pipeline. So when is this issue arising at all? What is being done that makes fork inefficiency anything other than academic interest. Must be something, right? Forking webserver?