Just because they are general and are relevant to almost every problem, doesn’t mean there aren’t traits you need to be aware of that might be relevant to your problem.

It’s like the Microsoft Word problem. 10% of functionality is used by everyone (eg saving a file, opening a file, changing the font). Those are baseline “everyone should know this”. The remaining 90% is still valuable and used by distinct subsets of customers. If Microsoft started deprecating the long tail of features without careful cohort analysis they’d start losing swathes of customers.

So for the most common operations of the most common structures I’d expect memorization of complexities just because of repetition of use (or deriving/guessing it from first principles cause it’s easy for the most common ones). But there’s much more traits and it would be important to understand that (eg what are the primary drivers of CPU time when accessing a hash map? An array? A linked list?).