For multi-sample runs (folder input), consider aggregating the metadata into a single TSV instead of one per sample. The output filename would then be just metadata.tsv, no longer prefixed with the sample ID. Would this be useful?