fix(bendpy): infer_schema does not support TSV format, breaking register_tsv()

[bohutang]

Summary

register_tsv() in PR #19444 calls infer_schema(..., file_format => 'TSV') to detect column names before creating the view. However, infer_schema only supports Parquet, CSV, and NDJSON formats.

In src/query/service/src/table_functions/infer_schema/infer_schema_table.rs, the read_data match is:

match info.file_format_params {
    FileFormatParams::Csv(_) | FileFormatParams::NdJson(_) => { /* ok */ }
    FileFormatParams::Parquet(_) => { /* ok */ }
    _ => {
        return Err(ErrorCode::BadArguments(
            "infer_schema is currently limited to format Parquet, CSV and NDJSON",
        ));
    }
}

TSV hits the _ branch and returns an error, so register_tsv() fails at registration time.

Suggested Fix

Add FileFormatParams::Tsv(_) to the CSV/NDJSON match arm. TSV is essentially CSV with a tab delimiter, so it should be able to reuse the same code path.

Related

• PR fix(bendpy): register_csv/register_tsv fails with column position error #19444
• Issue bendpy: register_csv() fails with 'Query from CSV file lacks column positions' #19443