Sizes a Model Can Read

A shop wants to feed T-shirt orders to a model, so the size column needs one-hot encoding. The sizes were typed by hand, so "Large", " large" and "LARGE" all appear, and some orders have no size at all (None).

Write encode_sizes(sizes), where sizes is a list with one entry per order.

1. Clean each size: remove spaces from both ends and make it lowercase.
2. One-hot encode the cleaned column with pd.get_dummies, using 1 and 0 for the flags. Name the column size before encoding, so the new columns are called things like size_large.
3. An order with no size gets 0 in every size column.

Return a dictionary with two keys:

• "columns": the list of new column names, in the order get_dummies gives them
• "rows": a list with one list of 1s and 0s per order, in the input order

At least one order has a size.