fix(python): Raise TypeError instead of AttributeError when calling next() on uninitialized GroupBy

[avasis-ai]

Summary

Calling next() on a GroupBy, RollingGroupBy, or DynamicGroupBy object without first iterating over it raises an unhelpful AttributeError: 'GroupBy' object has no attribute '_current_index'. This PR changes the error to a clear TypeError, consistent with standard Python behavior for non-iterators.

import polars as pl
next(pl.DataFrame().group_by(1))
# Before: AttributeError: 'GroupBy' object has no attribute '_current_index'
# After:  TypeError: 'GroupBy' object is not an iterator

Changes

Added a hasattr check at the start of __next__ in GroupBy, RollingGroupBy, and DynamicGroupBy to raise a descriptive TypeError when iteration state hasn't been initialized.

Closes #12868