For me, Linux was a hobby prior to and separately from programming. In the tutorials and documentation I read, every command was explained in detail and it was emphasized to me that I should never run a command that I don't fully understand. All instructions to elevate privileges were accompanied with advice about being careful running commands as root because root's privileges are especially dangerous. I was interested in those warnings, and took them seriously, because I wanted to master what I was learning. What I was learning, though, explicitly included Linux/Unix norms like security 'best practices'.

Developer documentation doesn't usually concern itself with Linux/Unix norms the way that tutorials for Linux hobbyists and novice sysadmins do. At the same time, the developers reading it might be perfectly dedicated to mastery, but just not really see what is considered proper usage by sysadmins (let alone the considerations that inform the boundaries of such propriety) to on-topic for what they're studying/exploring/playing with. Diving into those details might not be 'patt of the fun' for them.

What such a developer learns about sudo is mostly going to come from shallow pattern recognition: sudo is a button to slap when something doesn't work the first time, and maybe it has something to do with permissions.

But I think that comes from the mode of engagement, especially at the time of learning sudo, more than the mere frequency of use. I use sudo several times every day (including sometimes interactive sessions like you mention, with -i or -s), but I am careful to limit my usage to cases where it's really required. I'm not perfect about that; occasionally I run `sudo du` when `du` would suffice because I pulled something out of my shell history when I happened to be running it from / or whatever. But I certainly don't run it reflexively or thoughtlessly.