I disagree with @Air's answer. Datasets are a basic component of Data Science, there is no Data Science without datasets in the same way that there is no Data Science without Machine Learning. Should we then remove all questions related to Machine Learning and put it in Cross Validated? Should we remove questions related to optimization and add it to Mathematics. It is the same.
I think you are all forgotten the main purpose of this site, and maybe it is because SO don't specifically put it in the titles of the sites. The main purpose is to help people to solve their problems. All the rest is secondary.
If we don't allow to post questions about things strongly related to data science, then we are not helping. If we put many barriers to questions, people will be afraid to ask, we are not helping. If we get crazy about the location of questions or things like what is a perfect question, we are not helping.
Finally, if this place is not for helping people, then SO is really off-topic, and people will to other sites.
NOTE: Right now I'm not sure if I'm allowed to give an opinion (sustained with logical evidences). I might be wrong about this issue, and I know that somebody would argue that in another site there might be a higher change for answers (or maybe not), however, if the matter is deeply related I really think it is more important to focus on answering and helping.
I came across this article that explained very well the problem of SO as a community