The thing goes like this:
I wrote an experimental deflate recompressor that, much like precomp, does lossless recompression of zlib streams. It's much faster than precomp and can be a tiny bit stronger (because it supports multiple zlib versions and, I think, more zlib parameters).
It's experimental and done in a big rush, it's input is just a subset of gzip standard, there is no user interaction, barely any error handling, little documentation, likely bugs. I intended to bring it to usable state, but I can't get myself to start the work on it for a month already and I'm afraid I never will.
Nevertheless, it works, or at least mostly works and might serve as a reference or maybe even a base of something.
So on one hand I'd rather not release it because I'm quite ashamed of the sorry state of the code and on the other hand, I don't like it going waste.