Module names no longer clash with type and value names. The
tokenizer/parser still needs to be taught to be more careful in
identifying keywords, so that we can use 'str' and 'vec' and so as
module names.
* Cleans up the algorithm
* Move first pass to walk (second still folds)
* Support part of a type/value namespace split
(crate metadata and module indices still need to be taught about this)
* Remove a few blatant inefficiencies (import tables being recreated for
every lookup, most importantly)
* If an import was unused we would never print any errors for it.
* We would keep the existing environment in scope when descending 'foo.bar'
and would find 'bar' in the global environment if there was no 'bar' in
'foo'.
