fix(csv-parse): encoding detection with bom

[moroine]

This fixes adaltas/node-csv-parse#318

Describe the bug

Following test is failing:

import csvParse from 'csv-parse/lib/sync';

test('This is not working', () => {
      const data = Buffer.from('\uFEFFa,b,c\n"a","b","c"', 'utf16le');
      const records = csvParse(data, {
        bom: true,
      });
      expect(records).toEqual([
        ['a', 'b', 'c'],
        ['a', 'b', 'c'], // Got ["愀", "戀", "挀"]
      ]);
});

After some research, I found that while the options are re-evaluated upon encoding change, we still use old values for comment, escape & quote values.

[wdavidw]

Sounds good, thank you. It is now merged and published:

• csv-parse: 5.2.1 => 5.2.2
• csv: 6.1.4 => 6.1.5