To be perfectly realistic any kind of even relatively professional game is going to have tens or hundreds of thousands of lines of code in it, many files, many classes and functions.

You can pretty much always do something better, you can take some lines of code and manually write them in assembly and probably net better performance, or just rearrange the higher language code in order to make it more readable. But that takes time, time is valuable in coding, if we all had infinite time to work on every project I'm sure most of us would redo our code over and over to make it prettier and prettier.

I think you're making way too big a deal out of performance, because of that fact I just outlined above is -why- people make code to complete a task first, and then they optimize it if they need to. No use spending an hour rewriting one function out of 10k if its not even going to net you a visible time difference when the game is running.

That whole "performance obsessive" thing is usually what most coders go through when they're starting, then they grow out of it when they realize how wrong their priorities are.

EDIT: Oh and one more thing, the usual problem with code and performance is the same thing we've battled for a long time: the more generic something is, the worse it is at handling specific cases. If you make the most generic library in the world, the most specific library in the world will probably beat it by leaps and bounds.