Support configurable header row for CSV/XLSX sources

[SerylLns]

Problem

Some real-world files have metadata or description rows before the actual headers. For example:

(empty row)
*required, *required, *required, Ex: 45.398792, ...
public name, private name, address, GPS latitude, ...
ADRET, ADRET, COURCHEVEL 1550, ...

Currently, Sources::Csv and Sources::Xlsx always read the first row as headers (csv_content.lines.first / sheet.simple_rows.first). There is no way to skip leading rows.

Proposal

Add a header_row option (1-based index, default: 1) configurable per target or via import config:

Target DSL:

class HousingTarget < DataPorter::Target
  sources :csv, :xlsx
  header_row 3  # skip 2 description rows
end

Or via import config (runtime):

import.config = { "header_row" => 3 }

Both Sources::Csv#headers / #fetch and Sources::Xlsx#headers / #fetch would skip header_row - 1 rows before reading headers, and parse data rows from header_row + 1 onward.

Use case

Property management platforms where operators export templates with instruction rows above the actual column headers (e.g. *required, Ex: 45.398792). Asking users to manually clean files before import adds friction.