Noticed this while updating the polars implementation: their series must have a str name, and can't be None

We could hack around this in the implemenation, but I think it's easier to just require the name to only be str

Afterall:

• if a column is created from a dataframe (e.g. df.get_column_by_name('foo')), then its name will be the column name (which has to be str)
• if a column is constructed from a sequence or 1d array, then we can require that a str name be specified

So, name will always be str