Fixed csvformat's -H/--no-header-row functionality

Created by: dannguyen

csvformat's test_no_header_row() test seems to be incomplete/non-functional, e.g. it omits the default headers from the expected results:

    def test_no_header_row(self):
        self.assertLines(['--no-header-row', 'examples/no_header_row.csv'], [
            '1,2,3',
        ])

I changed the test to this:

    def test_no_header_row(self):
        self.assertLines(['--no-header-row', 'examples/no_header_row.csv'], [
            'a,b,c',
            '1,2,3',
        ])

And in csvformat.py, I copied the relevant snippet from cli.py:get_rows_and_column_names_and_column_ids() for generating default headers:

https://github.com/dannguyen/csvkit/commit/c4afb6cfb722eac0d7f6d806dc8af85a48251204#diff-5a5d6e33b5e9dc87dbe5b0db33e21e99R53

        reader = agate.csv.reader(self.skip_lines(), **self.reader_kwargs)
        writer = agate.csv.writer(self.output_file, **self.writer_kwargs)
 # new-code
        if self.args.no_header_row:
            # Peek at a row to get the number of columns.
            _row = next(reader)
            reader = itertools.chain([_row], reader)
            headers = make_default_headers(len(_row))
            writer.writerow(headers)
# /new-code
        writer.writerows(reader)