[MLton] flattening of arrays and vectors

[Stephen Weeks]

> With regards to the September release, is that version planned to
> have the undumped (save world) processing of the basis library?  Now
> that MLton can be used during development, it would be very nice to
> have the time from start to detection of type/syntax errors be
> shorter.

I agree with Henry that this would be a nice feature.  I am thinking
that something really simple might give a lot.  Matthew, do you have
any thoughts on how hard this would be?