Python is fairly simple in implementation while being as flexible as possible.
Look up the concept of descriptors for something fairly difficult to optimize when you consider __getattr__ or __getattribute__. Consider also that you can directly access the __dict__ of both an instance and of the class of that instance. You can monkey-patch the class extensively or directly into it's __dict__. You can even replace the __class__ reference on instance to make it a subtype of different class (which is insanely flexible). Then there's the stuff about metaclasses, which is fairly advanced and collides with all of the other flexibility.
Basically pretty much anything not from __builtin__ or from a C extension is entirely mutable at any time. Consider one of the problems PyPy had with some packages. PyPy didn't support `locals()[42] = value`, which some packages actually depended upon. Nor does PyPy support `instance.__dict__[42] = value` which even more depended upon. Generally Python used a lot of simple concepts such as mutable dictionaries with heterogeneously-typed keys as the implementation of how classes, modules, etc works, which is flexible but very hard to optimize in any way (especially since they are mutable).