Support of CSVY would add no new dependencies, so it could be part of Cartmandu core. The CSVY importer could be used as general CSV importer as well if we assume no normal CSV file matches /^---\s*$/. If so, buffer all lines until /^(---|\.\.\.)\s*$/ or EOF and use this as YAML header, read by YAML importer. The rest is normal CSV. The metadata could be used for validating as table schema (see #207) but also as independent document. Note that the YAML header contains order of fields which is not preserved by normal CSV importer! The CSV importer could automatically generate a fields field to be used for exporting tabular format with fields in the same order.