[racket] TR: struct: #:prefab option

On 2013-05-06 13:52:57 -0400, Asumu Takikawa wrote:
> Whoops, Vincent pointed out that my claim wasn't quite right. Here, this
> is just a type->contract bug, but it would be prevented if the right
> contract were applied to `foo-f`.

Err, this was the wrong explanation too. Actually neither of the
mutation examples are a fundamental problem, because an appropriate
`struct/dc` would protect the read appropriately. It doesn't matter what
the untyped side mutates as long as you protect it on the read.

Cheers,
Asumu