Because most people who use Python use it not because of the language itself but because of its library / framework eco-system and PyPy's compatibility with that is limited.
Also PyPy offers no performance advantage if all the CPU intense stuff happens in libraries written in C/FORTRAN/ASM anyway.
I think they should really make the project in sync with cpython. Because already they have a huge gain, for pythonic stuff.
In many cases people don't care about performance (CPU time) but they do care about problems, real or imagined, if they try something new. (Human time)