limit check insertion

[Stephen Weeks]

> Thinking about this a little more, I don't think it's a bad idea to move
> teh GC loop into the runtime system.  On the other hand, it's not going to
> save any code.  Looking at the "in assembly we would have" below, this is
> identical to what is currently there, except that the jmp skipGC is
> currently jmp stackCheck. 

Makes sense.  This reminds me, is there a bug in the ccodegen.h since
it doesn't do the stack check in the loop?  I would think so, since
the gc might have shrunk a thread's stack while it's swapped out.
I think that the stack check should be part of the loop like the limit
check is.