[FEA] Support `skip_rows_after_header` in CSV reader options

**Is your feature request related to a problem? Please describe.**
In cudf-polars, we want to read the header to construct the schema, but still skip rows immediately after it. Currently, we cannot do this without a separate call to fetch just the header line.

**Describe the solution you'd like**
A new CSV reader option: `skip_rows_after_header`, which specifies how many rows to skip after reading the header (if present). This allows for reading the header row while skipping over empty or irrelevant rows following it.

**Describe alternatives you've considered**
In cudf-polars, we currently implement a workaround by reading just the header in a separate python call before constructing the full CSV scan.