Why do you want this feature?

Other packages (like Pandas) that provide a read_csv function have this option names
Which allows you to override the column names, even if the file has a header.

We can currently provide:

• columns as a struct, containing the name + type
• column_types or types or dtypes as a list or struct, containing the (optionally partial) types for the columns

But we have no way of setting only the names

Maybe taking the same approach as the solution for the types is best here
providing a way to set names for the columns, either partially (sequentially) with ['name1', 'name2', 'name3'] or as a struct, from old_name -> new_name {'old1' : 'new1'}