[extractor] Use classmethod/property where possible

and refactor lazy extractors accordingly.

This reduces the need to create extractor instances

yt_dlp/extractor/youtube.py

@@ -31,6 +31,7 @@ from ..utils import (
     NO_DEFAULT,
     ExtractorError,
     bug_reports_message,
+    classproperty,
     clean_html,
     datetime_from_str,
     dict_get,
@@ -5781,16 +5782,17 @@ class YoutubeMusicSearchURLIE(YoutubeTabBaseInfoExtractor):
 class YoutubeFeedsInfoExtractor(InfoExtractor):
     """
     Base class for feed extractors
-    Subclasses must define the _FEED_NAME property.
+    Subclasses must re-define the _FEED_NAME property.
     """
     _LOGIN_REQUIRED = True
+    _FEED_NAME = 'feeds'
 
     def _real_initialize(self):
         YoutubeBaseInfoExtractor._check_login_required(self)
 
-    @property
+    @classproperty
     def IE_NAME(self):
-        return 'youtube:%s' % self._FEED_NAME
+        return f'youtube:{self._FEED_NAME}'
 
     def _real_extract(self, url):
         return self.url_result(