[racket] TR: predicate for a Float ?

At Fri, 6 Jan 2012 12:10:58 -0500 (EST),
J. Ian Johnson wrote:
> All unions get flattened, so generally there are multiple possible
> reconstructions of larger types. It's a set-cover problem -
> NP-complete. But hey, proving well-typedness in TR is co-NPC,
> so... heh. I think Vincent has a greedy algorithm to do a good job of
> this, but apparently it hasn't been merged into the main
> branch. What's the status of that, Vincent?

It's been on master for a while.

Vincent