[CM] Set C object's pointer value

[bil at ccrma.Stanford.EDU]

This should be safe if it isn't called in an
iterative context that might be optimized.
If called there (that is, in scheme via some
wrapper function), make sure that function is
not considered "safe", so the optimizer will
give up when it is encountered.  I'll add
the s7_c_object_set_value to s7 -- thanks!
As far as I can tell, c_pointers aren't
optimized, so it looks like they can safely be
changed.  I'll write some tests of both cases
to see if I've missed something.