Deep twin issues

[Paolo Redaelli]

I'm working on the deep twin bug.

At first I thought it was related to the C runtime.

Then I realized that the issue related to the signature of the natively 
stored string have at least another potential cause: queries have side 
effects.

So I started first removing those side effects from queries and moving 
them into commands.

It's only a start. In fact a quick run of "se test" into 
test/language/memory showed that there are further issues related to C 
run time that deals with deep_twin.

I'm investigating it, but I fear or better I suspect that it may be more 
feasible to reimplement deep_twin using introspection.

I will keep you informed

    Paolo