A Split You Can Repeat

Before training a model, data is split into a training part and a test part. The split is random — but it must come out the same on every run.

Task: write split_data(items, seed) that returns a list [train, test].

1. Call random.seed(seed) once, at the start.
2. Go through items in order. For each item, call random.randint(1, 10) exactly once.
3. If the number is 8 or less, the item goes into train; otherwise it goes into test.
4. Both lists keep the items in their original order.