Parsing with `columns: true, cast: true` throws for csv data that has numeric column headers

[chriswheeldon]

Describe the bug

Parsing with columns: true, cast: true throws for csv data that has numeric column headers.

The following error is, for example, thrown in parse:

CsvError Error: Invalid column definition: expect a string or a literal object, got 2023 at position 1

I think ideally columns: true, cast: true would not attempt to cast header row elements and instead preserve them as strings. I'd be happy to contribute a fix if that makes sense as a solution!

Note, adding from: 2 does not workaround the issue. The only workaround I could find was to implement a cast function like,

cast: (value, context) => {
  if (context.header) {
    return value;
  }
  const casted = Number(value);
  return isNaN(casted) ? value : casted;
},

I am using csv version 6.4.1.

To Reproduce

import dedent from "dedent";

// Dataset with numeric header elements and numeric values
const data = dedent`
  Person,2023,2024,2025
  John,100,150,200`;

const parsed = parse(data, {
  columns: true,
  cast: true,
}); // THROWS