IntInf asserts

[Matthew Fluet]

> > Could also help with threads.  
> ...
> > I don't know if it would buy us much besides a potentially cleaner
> > interface; the primitives would no longer need to be considered
> > entersRuntime, and a straight C call is slightly cheaper than a
> > runtime call.
> 
> If it works out, that sounds like a win to me, mostly on the
> cleanliness side.

One minor point I forgot: in order for the copied thread to be runable, it
needs to have been copied during an entersRuntime -- because we need a
return address at the top of the stack.  Still, we wouldn't need to do the
limit check in the primitive itself.