The first rule of optimization is never to code what you "know" is efficient. Code it to work, then measure where it needs tuning -- if anywhere. I've been relied on to provide efficient, reliable software since before we herded velociraptors here on the Ranch because cows hadn't been invented yet. And every blamed time, the place where the performance problems were found were not the places that people "knew" there would be inefficiency.
Obviously, using a bubble sort where a heapsort is appropriate is a type of early optimization that should be considered. Although I had one major production system where the data was in worst-case order for heap- and quick- sorts and the optimal was a Shellsort, instead.
But worrying about imagined overhead for trivial common functions is a waste of your time and effort. Until proven otherwise. Anything really bad would have been optimized by the framework designers, since they, too, have a stake in efficient software.
An IDE is no substitute for an Intelligent Developer.