I agree, but... One problem is that we are maxed out on custom close reasons. We already have a bit of a problem with suboptimal communication of the "language specific" reason, so it would be a good thing to be clear about what's okay and what's not okay. In fact, it would be manifestly unfair to ban questions from people who are creating constructed languages, instead we should ban certain kinds of questions. Here are some example questions (with commentary).
1: "I'm working on a conlang. How frequently do SOV languages have AN word order?" (The second sentence is a legitimate question: does the first assertion poison the question?)
2: "I want my conlang to have VOS word order and case marking. Should the case marking be a prefix or a suffix?" (I think this ought to even be off topic for a conlang site, since there's no basis for holding either "yes" or "no" to be correct – pure opinion).
3: "Would there be any problem if I build my conlang so that it is uttered during inhalation rather than exhalation?" (A more clever questioner would have asked "are there any languages that are uttered during inhalation rather than exhalation, and if not, why not?", and could have applied that to their conlang in private).
4: "I constructed a conlang 5 years ago, and there are a couple dozen speakers. I'm thinking of adding features to the language. What do you think of these additions..." (I'm taking advantage of the fact that there are speakers, which I might think licenses asking any questions at all).
I would suggest that all questions about conlangs are off topic, except factual questions about the structure of an existing conlang (with speakers). Though except for Esperanto and Klingon, I don't know what conlangs have speakers, so it's not clear that the "with speaker" criterion is useful. I think the restriction should be about there being a way to actually verify an answer, such as a published grammar.
I disagree with the disagreement (not trying to encourage an infinite regress). Put very simply, there is no reasonable and answerable question about conlangs that cannot be framed strictly as a question about natural languages. If you want to decide whether to make your conlang VOS, you should decide privately how that question could ever be answered objectively. If you want the language to exhibit the least-frequent properties of human languages (or the most frequent), then ask a question about natural language. Don't ask for opinions about what you should do.
I also disagree with the position that linguistics is specifically the study of any given language. Linguistics is the study of the nature of language, and not the accumulation of studies of individual languages. Linguistics is not the social history of the word "meat" in English, it is not about normative judgments such as the prohibition against "ain't", it is not about programming tips for text mining, and it is not answers to questions like "What is the accusative plural of 'horse' in Sanskrit?".
The implication that we are trying to impose a world view on innocent people is wrong ("people have an agenda and try to impose it on others as though they're not allowed to ask questions here"). There is a huge difference between a focused Q&A website, and an open ended discussion forum. If you want an open-ended discussion where all comments are equal, then go to a discussion forum, and not a fact-seeking site. The very reason why SE exists at all is that there are technical questions which need answering by knowledgeable people. The rules of SE are there to preserve the function of SE, to keep it from turning into a discussion forum which just accumulates personal opinions. Of course there are questions that you are not supposed to ask here.