[MLton] type error messages

[Matthew Fluet]

> No matter what we do there will always be room for human explanation
> of type errors.  I think it would be nice to add a TypeErrors page to
> the wiki with a catalog of all type error messages along with example
> programs that generate them.

Thanks for the info.  Also, I'll point out that we were generating
specifically convoluted examples (excercising the Definition), so it's not
clear that these are likely to come up in practice.

> It would also be nice to add some of the above explanation to the
> Elaborate page on the wiki, as well as some of the more user-oriented
> stuff to the TypeChecking page.  People shouldn't have to infer the
> closest-definition rule.  If you understand how things work, the error
> messages should be clearer.

Yes, updating the docs is a good idea.  And, I recalled enough of the
emails about the the type-checker to quickly jump to the
closest-definition rule.  Although, it wasn't until we applied that rule
to everything that some of the error messages made sense.