Python on mobile systems

It is possible to get a fully self-contained, static Python build (we do it and even include NumPy using some truly evil hackery to deal with NumPy's nasty build setup). There is a patch required though to reconcile some problems with built-in and frozen modules under the same module path, but it is minor (and it is Python2, so no hope of it getting applied).

Basically, we use a CMake build instead to build the required modules, optionally builtin, with a flag for freezing the Python code. Then, we build NumPy statically after running its build to grab its configured sources. The application then injects NumPy into the built-in tables just like its own modules. Not pretty, but when you need to support machines without shared library support and slow disks (supercomputers), there's not much that will be pretty in the end.