Support auto-splits for file datasets

[PGijsbers]

Listing TODOs on the issue tracker (#172).

automlbenchmark/amlb/datasets/file.py

Lines 343 to 352 in 98bf554

	class ArffDataset(FileDataset):
	def __init__(self, train_path, test_path, target=None, features=None, type=None):
	# todo: handle auto-split (if test_path is None): requires loading the training set, split, save
	super().__init__(
	ArffDatasplit(self, train_path),
	ArffDatasplit(self, test_path),
	target=target,
	features=features,
	type=type,
	)

automlbenchmark/amlb/datasets/file.py

Line 425 in 98bf554

# todo: handle auto-split (if test_path is None): requires loading the training set, split, save

Given the file with all data, and the target column, it should be trivial to support calls where the test split is not yet a file for ARFF and CSV datasets.