We don't technically have a policy on this. I'm of the opinion that it should be discouraged when possible, and should not be encouraged at any point. This is how we handle our objective lists, after all - when someone asks for a list of weapons, we don't list each weapon individually.
One-per-answer is not bad because of game-rec. Nor is game-rec bad because of one-per-answer. It's more of a coincidental issue.
This is an old post, but I'd like to point to a Cooking post by Aaronut which reflects some of the thoughts about one-per-answer versus other forms of list building. In essence, it explains the different approaches to lists, and some basic concepts of why one-per-answer is something to not be encouraged and ideally should be discouraged. I will quote some bullet points still relevant regarding the issues with one-per-answer (the final is less an issue now that CW has been changed).
• It unconsciously ascribes ulterior motives to the question author, which is unfair;
• It does not prevent badge farming, which is still a problem;
• It ignores more conventional incentives for over-participation - socializing and herding;
• Voting patterns suffer from chronological bias (oldest answers are typically favoured);
• Pagination makes a large number of answers incredibly difficult to read;
• Once a certain momentum is reached, answers become heavily duplicated due to the above;
Badge farming isn't as much an issue in my opinion, but the other points are some of the large issues with it. Namely, that it leads to a lot of answers which has issues with duplication and readability, and that voting is drastically different. I'd like to add two more points to it.
It conflicts with the otherwise solveable nature of a question. One big thing about the site is the availability of an accepted answer. When your answer pool consists of a bunch of equally correct answers that really have no reason to have one accepted over another (as a list of useful apps may well be), then it leaves us with an eternally open question.
Voting is on the scale of the whole site, not individual questions, but one-per-answer encourages treatment like the other. When you think of it conceptually, one-per-answer sounds cool because users can rank them in relation to each other. What people always forget is that votes are measured for the entire site - if something is rated a 34 to put it at the top of the list, it also ends up dwarfing 70% of the objectively-voted answers on the site.
One can mitigate many of these issues by having one large answer that contains all of the options instead of individual ones. This gives you something to authoritatively pin to the top of the question, and also serves to help keep things on one page and avoid duplication. Moreover, this means that when you see a list that gets 30 upvotes, it is more likely because it's actually a good list, rather than it is a random app that people happen to like.
Whether the big answer is Community Wiki or not depends on the question's needs. If someone can build a comprehensive list on their own, which is more likely with finite lists, then Community Wiki is largely unnecessary. But if it's something that needs everyone to build on, as most resource lists would be, then it would be better off as Community Wiki to aid in that endeavor.