(TL;DR at bottom.)
I think there are a few different aspects to this, and these different aspects don't all need to be treated in the same way. I want to break this question out into the following:
- Close reasons
- Specific examples
- Repertoire lists
Close Reasons
In my opinion, the prescribed collection of close reasons are all just simplistic attempts to codify the same thing while simultaneously offering constructive feedback to the asker.
The close reasons most relevant to this discussion usually amount to "not about music", "too broad", or "too localized". Obviously on Music.SE we deal with "too localized" questions in a much different context from the programming-related StackOverflow on which the principle was designed, but the basics still apply. These are:
- Questions at their core should be about a discrete problem you are trying to solve
- These problems should theoretically be applicable to another person at a different time with different circumstances.
The objective of these principals is to keep StackExchange more like a Q+A site than a forum, and that all user contributions (answers and questions) contribute to a perpetually useful knowledge base for the benefit of all.
Questions that need to be closed usually fit within one of two categories:
- Questions that have musical problems, but are not being asked in a manner that addresses the issue at hand.
- Questions that look like musical problems, but once you get to the root of the issue at hand, they are actually about something different.
In the second category, often it's impossible for someone to realize this without musical training, and as such, comments allow us to point the asker in the right direction as we send them on their way. For questions in the first category, the whole purpose of renaming [closed]
to [on hold]
was to communicate that the question could be a good one, but it needs to be revised or clarified by the asker before being reopened.
This is also where most of the close reasons come into play, and why they tend to amount either to "too broad" or "too localized". For example:
Let's say you are an inexperienced choir director and you need some help programming a spring concert. Some lousy questions you could ask in this situation might be "What music should I program for a middle school choir concert?" or "What piece should I use to balance a choir program consisting of piece x and piece y?"
The first question would be closed as "too broad" since it is completely open-ended, even if it was "well-defined" as to include the choir's current ability level and each child's range. (Being well-defined does not mean the question is a good one.) The second question would be closed as "too localized", since it would be useful only to you at this particular moment in time. The right question to ask might be something like "What aspects of repertoire should I consider when programming a concert for a middle school choir?" This question sits right in the sweet spot, on top of the problem at hand, and a really great answer will probably include some examples of repertoire you might be able to use, or some resources for you to peruse, while still being useful for other people who have the same problem in the future.
In light of this and my first point about close reasons being simplistic, I'd very much agree that the "song identification" criteria can be applied too broadly and in a non-constructive way. The "off-topic" close reasons that we have access to can be edited by a moderator, but we only have a default reason and three custom ones to choose from, so we have to choose based on the highest-volume types of closes that arise. Of course, this is also open to discussion on Meta, and if our high-rep users decide that different reasons would be more appropriate, we will change accordingly.
As it exists now, the "song identification" close reason is meant to be used for questions where the asker provides a description or recording of some music and expects an answer of how to find it or music like it on record. These questions typically come from brand new users with little musical training, and they occur very frequently. Furthermore, the questions hold almost no value to future readers, and the only way to answer such a question is to guess properly. The community has been pretty clear that this is not what they want the site to be about, so the close reason exists as it does.
Aside: In researching this answer, I came to the realization that our reputation limit for the privilege of casting close votes is still set to that of a day 1 beta site, even after THREE YEARS in Beta. (Incidentally, happy belated birthday to Music.SE!) This has not scaled up with the general reputation of our user base, and so many more people have the ability to cast a close vote than would typically be the case were we out of beta (500 rep instead of 3,000). I think this is something that myself and the other mods should consider when reviewing questions closed by non-moderator vote.
Specific Examples
I'd say of your three examples, this one has the most merit: I think it is potentially useful to future readers, and the question itself was well-asked and well-answered. There are some valid points for closing; the root of the question really isn't about music so much as it is "How do I find and navigate Mozart's catalogue?" But, I would agree that the "song identification" criteria was incorrectly applied here. Personally, and in light of the low-rep votes that caused the question close, I would like to see this question edited to remove the limitation of excluding arrangements, nominated for reopening and made community wiki--more on that later.
In this case, while the subject matter is more interesting than the typical song identification question, it is pretty clear to me that the asker heard a single example that he became fascinated with, and is now coming to us to try to find more music like it (the asker asked the same question in the comments of the YouTube video he linked). In other words, a textbook example of a song identification question. The reasons I detailed above apply to this, and I agree with the close. There are certainly lots of good questions that could be asked about retrograde or inverted melodies, and those answers could conceivably (I hope) include examples, but this question is pretty clearly off-topic according to the site's criteria.
This question, to me, is very confusing. The close reason probably should have been "Unclear what you're asking", since it is unclear what problem the asker is attempting to solve with the question--instead, it just seems like academic trivia. There's nothing wrong with academic trivia, of course, but that is more suited to a forum (or chat) than it is a StackExchange Q+A site. Again, there are certainly interesting questions that can be asked about vocal harmonization in popular music!
I want to also give a reminder here that close votes and edits go both ways -- none of the questions that were linked got any "reopen" votes, nor were any of them edited after the close to improve them. Any user able to cast a close vote (reputation >500, at present) should be able to cast a reopen vote. (I don't believe there are any systematic things I'm missing that prevent this, but correct me if I'm wrong.) It's far more common that closed questions simply need to be improved than it is that the site's definition needs to be modified.
Repertoire Lists
This has been on my mind for some time as an aspect of the site's scope that I think we should modify. Generally, questions looking for repertoire recommendations tend to be closed as "too localized" or "opinion-based", but I think a case can be made for accepting certain questions like this in our subject area. Not every site needs to be run exactly like StackOverflow (certainly, many other sites have deviated significantly), and I think we should try to take into account the nature of classical music when thinking about these questions. What I mean by this is that repertoire moves VERY slowly in the classical and academic worlds, and questions about what that repertoire is are both long-lasting and objectively answerable. If I speak to any other classically-trained trombone player in the world, the words "Grøndahl" or "David" (provided you pronounce them properly) mean exactly the same thing -- a trombone concerto that everyone has either played or heard in studio class dozens of times. To this end, I would love to see us adapt the Community Wiki feature. This would allow us to keep useful lists of important repertoire itemized by popularity, but prevent undesired gains in reputation resulting from the simple contribution of the name of a piece.
Conclusion/Short Version
I hope this sheds some light on what we are trying to make Music.StackExchange into, and why. Examples and responses to various questions are included above, but it is important for us mods to keep an eye out for questions that are closed by low-rep close votes. At the same time, users should be proactive in figuring out how to edit, improve, and reopen questions so that meta discussions can be more specific and efficient instead of long-winded and overarching. Finally, I would like to see Music.SE embrace Community Wiki for repertoire list questions that are well-asked, objectively answerable, and perpetually relevant. Looking forward to seeing what people have to say about this in the comments!