Fix imports and general cleanup

· Import from compat what comes from compat. Yes, some names are available in utils too, but that's an implementation detail.
· Use _match_id consistently whenever possible
· Fix some outdated tests
· Use consistent valid URL (always match the whole protocol, no ^ at start required)
· Use modern test definitions

youtube_dl/extractor/tapely.py

@@ -4,10 +4,12 @@ from __future__ import unicode_literals
 import re
 
 from .common import InfoExtractor
-from ..utils import (
-    ExtractorError,
-    clean_html,
+from ..compat import (
     compat_urllib_request,
+)
+from ..utils import (
+    clean_html,
+    ExtractorError,
     float_or_none,
     parse_iso8601,
 )