feat(rust, python): add truncate_ragged_lines

[ritchie46]

closes #10585

[Wainberg]

I raised the initial issue (#10585). I really like how this pull request causes an error to be raised when reading a ragged CSV, instead of silently truncating the ragged columns.

However, I disagree with the need for a truncate_ragged_lines argument, because it doesn't seem like a common use case to want to lop off part of your file before reading it.

Instead, I'd suggest allowing the user to set new_columns to be longer than the number of columns in the first line. This would make polars consistent with pandas, as illustrated here.

Then, the error message that's raised when reading a ragged CSV could recommend one of two solutions:

1. setting infer_schema_length=None to parse ragged data with no issues, at a potentially large runtime cost; or
2. setting new_columns to be as long as needed to encompass all the ragged columns.

[ritchie46]

However, I disagree with the need for a truncate_ragged_lines argument, because it doesn't seem like a common use case to want to lop off part of your file before reading it.

You don't have to use it

Instead, I'd suggest allowing the user to set new_columns to be longer than the number of columns in the first line. This would make polars consistent with pandas, as illustrated #10585 (comment).

Polars needs a schema. New columns is a rename after the fact. The argument is only for renaming columns in the csv-file. I don't want this control anything else.