Fix bug in load_from_files_multilabel()

When loading documents from a folder, the order in which they initially
appear in the ``labels_path`` file wasn't kept because a dictionary was
used to construct the (document name, labels) pairs. This issue
sometimes caused the pytest failing in line 252 of the "test_util.py"
file.

pyss3/util.py

@@ -1669,15 +1669,17 @@ def load_from_files_multilabel(docs_path, labels_path, sep_label=None, sep_doc='
                 doc_labels_raw = [re.split(sep_label, l.rstrip())
                                   for l in flabels.read().split('\n')]
                 doc_labels = {}
+                doc_names = []
 
                 for doc_name, label in doc_labels_raw:
                     if doc_name not in doc_labels:
                         doc_labels[doc_name] = [label]
+                        doc_names.append(doc_name)
                     else:
                         doc_labels[doc_name].append(label)
                     cat_info[label] += 1
 
-                for doc_name in tqdm(doc_labels, desc="Loading documents"):
+                for doc_name in tqdm(doc_names, desc="Loading documents"):
                     file_name = doc_name + ".txt" if '.' not in doc_name else doc_name
                     with open(path.join(docs_path, file_name), "r", encoding=ENCODING) as fdoc:
                         x_data.append(fdoc.read())