[cairo] cacheable image representation now on a branch

[Paolo Bonzini]

>> 1) I fixed the large-source failure for Quartz.
> 
> I don't if modifying the test is the best way to fix the problem.
> Instead it probably makes sense to have api for querying the maximum
> size of a surface à la glGetIntegerv(GL_MAX_TEXTURE_SIZE, &texSize);

What about also (simpler to do) a CAIRO_STATUS_SURFACE_TOO_BIG error and
halving the size in the test until the error does not show anymore?

Paolo