[ENH] set mirrors for time series classification data loaders (#5260)

This sets the mirrors for time series classification data loaders:

1. to the new URL of the UEA repository (the one that is constantly
changing)
2. to the mirror URL on the sktime GitHub, as a backup

Testing of the download utility is re-enabled.

Related: #4754, #4749

sktime/datasets/_data_io.py

@@ -233,7 +233,10 @@ def _mkdir_if_not_exist(*path):
     return full_path
 
 
-CLASSIF_URLS = ["https://timeseriesclassification.com/ClassificationDownloads"]
+CLASSIF_URLS = [
+    "https://timeseriesclassification.com/aeon-toolkit",  # main mirror (UEA)
+    "https://github.com/sktime/sktime-datasets/raw/main/TSC",  # backup mirror (sktime)
+]
 
 
 def _load_dataset(name, split, return_X_y, return_type=None, extract_path=None):
@@ -290,12 +293,15 @@ def _get_data_from(path):
     if extract_path is None:
         extract_path = os.path.join(MODULE, "local_data")
 
+    # in either case below, we need to ensure the directory exists
+    _mkdir_if_not_exist(extract_path)
+
+    # search if the dataset is already in the extract path after download
     if name in _list_available_datasets(extract_path):
         return _get_data_from(extract_path)
 
     # now we know the dataset is not in the download/cache path
     # so we need to download it
-    _mkdir_if_not_exist(extract_path)
 
     # download the dataset from CLASSIF_URLS
     # will try multiple mirrors if necessary

sktime/datasets/tests/test_single_problem_loaders.py

@@ -81,9 +81,6 @@ def test_load_numpy2d_multivariate_raises(loader):
         X, y = loader(return_type="numpy2d")
 
 
-@pytest.mark.xfail(
-    reason="repeated upstream location failures, see 4754. xfail until fixed."
-)
 def test_load_UEA():
     """Test loading of a random subset of the UEA data, to check API."""
     from sktime.datasets.tsc_dataset_names import multivariate, univariate