There's already a common way to add metadata to csv : use the first row. It is used for column names, but we can extend it to add any metadata. It would be easy

There's already a common way to add metadata to csv : use the first row. It is used for column names, but we can extend it to add any metadata. It would be easy to implement in current parsers and it would fail in a, if not graceful, predictable way probably already accounted for (i.e. garbage in the first row)