QOpenGLBuffer::allocate/write uses int instead of a 64-bit type

src/opengl/qopenglbuffer.h says

    bool read(int offset, void *data, int count);
    void write(int offset, const void *data, int count);

    void allocate(const void *data, int count);
    inline void allocate(int count) { allocate(nullptr, count); }

which limits all this to 32 bits. However the underlying API does use a 64-bit aware type:

void glBufferData(GLenum target, qopengl_GLsizeiptr size, const void* data, GLenum usage);

This would be trivial to fix if it were not for SC/BC constraints
Changing the type in the existing methods would be BIC.
Adding overloads with the same name but different types would 1) have to be skipped on 32-bit platforms, 2) maybe create ambiguity in some cases? I tried calling `foo(int)` and `foo(long)` with a `short` and it wasn't ambiguous though, so maybe this would actually work?