[Ecls-list] unboxed closure

[Brian Spilsbury]

Well, I have unboxed locals and globals seeming to work reasonably well.

The main remaining cause of conversion seems to be the cl_object *lex
vector used by closures.

I'm thinking about converting this to a vector of unions of
representation types, and since the closure knows which slot is what
at compile-time, it should be able to say lex0[0].cl_fixnum or
lex0[0].cl_object and so on.

I guess this would make closures more efficient than structures.

Anyhow, if anyone can see anything terrible about this idea, let me know.

Regards,
Brian.