You have made clear to me why my attempt for a RFE for COPY FROM CVS has found some technical resistance/disagreement, but I still think my idea even if not so popular for concrete and cultural reasons makes at least sense to some people
It's a perfectly reasonable problem to want to solve; the question is whether COPY is the right place to solve it. I would think that a tool that reads the CSV data and produces a proposed schema definition for the table would be a more generally-useful approach.
