feat: expose scene presets as voice-controllable scene.* entities (0.5.0)

Add a SCENE platform (scene.py) so each scene preset becomes a scene.*
entity. Alexa / Google Assistant / HomeKit expose the scene, light, and
switch domains — but NOT button — so the existing button.* scene entities
are invisible to voice assistants. Scenes use a distinct unique_id so both
coexist (buttons for dashboards, scenes for voice).

- Register Platform.SCENE; bump manifest 0.4.0 -> 0.5.0
- README: "Voice Control (Alexa / Google / Siri)" guide (Nabu Casa cloud,
  manual skill linking, HomeKit bridge, voice-friendly naming)

custom_components/ledgrab/manifest.json

@@ -8,5 +8,5 @@
   "iot_class": "local_push",
   "issue_tracker": "https://git.dolgolyov-family.by/alexei.dolgolyov/ledgrab-haos-integration/issues",
   "requirements": ["aiohttp>=3.9.0"],
-  "version": "0.4.0"
+  "version": "0.5.0"
 }