I wouldn't be surprised if this has been extensively discussed in the past. I wonder if it would not be possible to correct and refine the taxonomy of questions which should be closed. The possible reasons: "duplicate", "unclear what is being asked", and "too broad" are OK and deserve their independent places on the list. Then there is what appears to be the grab-bag reason "lacking context...". This is more or less a vague way of saying "just plain low quality". First of all, it doesn't belong under "off-topic". Clearly many of the questions closed for this reason are perfectly clear mathematical questions, which are on-topic but nevertheless bad questions for a number of possibly quite different reasons. For example, they might even be good exercises for some mathematical subject, and clearly on topic, but pretty obvious, and their being asked means that the asker hasn't learned to sit and think. Could someone more involved in the site, perhaps by making a survey of bad questions, propose a better taxonomy of unacceptable clutter, so that one would have better boxes to check off, i.e. better criteria for closing.
Edit: I’ll summarize what I have learned from the discussion.
1) Custom reasons have to be under “off-topic” and we only get maximum 3 custom reasons.
2) We already have 2 custom reasons, namely “lacking context or details” and “seeking personal advice”.
3) It probably would be rather difficult to reach a consensus about a third custom reason (and I have to say I wouldn’t even be able to formulate a proposal).
4) We already have the possibility of using the “off-topic/other” reason, with a comment, and when we do the comment gets placed as a comment to the original question, which has the form "I'm voting to close this question as off-topic because…"
My own suggestion would be to make more frequent use of the “off-topic/other” category when “off-topic/lacking context..” doesn’t quite fit, or in order to give more specific feedback. The likely outcome will be that anyone who does this regularly will eventually have his/her own small collection of useful boilerplate responses.
Thanks to everyone who gave some feedback.