I think code review should be considered on-topic. This platform is well equipped for posts with potentially multiple longer answers and without the concept of an accepted answer, they can very well deal with different aspects of the code.

I also think this is a good place, because we already have here a community of very knowledgeable users who can provide helpful answers which will not only help the OP but also other people who might read about things like best-practises. I always found it strange that some other Q&A network pushed of code reviews into a separate community with much less experts (at least in the tex tag).

It seems very reasonable to me to allow users here to discuss codes. On the TeX.SX sites there are quite a few posts in which users post excerpts of a library and ask, say, details about `\pgfdeclareshape`. I think that this site is much more suitable for this task. First of all, as @samcarter already mentioned the (IMHO really stupid) accept-an-answer "feature" is not around. Second, I feel that the comment column here is much better equipped to get going. After all, one can post figures and it is more intuitive to have things side by side.

Overall I am very supportive of this proposal, and would not be surprised if some packages get improved or even created this way.

I think we can provide this service. But before that I'd like a feature to be implemented for this type, without which I'd find code review tedious:

- code lines
- some sort of (hidden) versioning for those code lines (if the question is edited, I'd like to have the answer still make sense)

Imho, the code review tools of GitHub and similar software are excellent (opening discussions for single code lines or blocks of lines). I'm not sure whether this should or would be implemented in a format like TA, if it would, I'd be happy to support code reviews in general. If TA came up with something even better, that'd be really great (but I have no idea on what could be provided).