feat: user-customizable event voice rules (gift/blind box/enter/follow/like/guard/super chat)

- New event_rules.py: load/save event_rules.json, robust template renderer
  (missing vars -> empty, empty template -> muted), guard level mapping
- danmaku_handler.py: custom SEND_GIFT callback to extract blind_gift_name;
  implements all event callbacks feeding the TTS queue per user rules
- server.py: GET/POST /api/rules, load rules at startup, pass to client
- Web UI: 'Event Voice Rules' panel (per-event toggle + template + save)
- main.py CLI loads rules; build.py bundles event_rules.json
- skip_free_gift option to mute silver free gifts
- Rebuild portable zip (83.8 MB)

main.py

@@ -13,6 +13,7 @@ import sys
 
 from danmaku_handler import DanmakuClient
 from chinese2kana import chinese_to_kana, filter_kana
+from event_rules import load_rules
 from tts import TTSBridge
 from audio_player import play_wav_async, get_output_devices
 
@@ -71,9 +72,11 @@ async def main_async(args: argparse.Namespace) -> int:
     queue: asyncio.Queue = asyncio.Queue(maxsize=256)
     shutdown_event = asyncio.Event()
     bridge = TTSBridge(voice=args.voice, speed=args.speed)
+    rules = load_rules()
     danmaku_client = DanmakuClient(
         room_id=args.room_id, queue=queue,
         cookies={"SESSDATA": args.sessdata} if args.sessdata else None,
+        rules=rules,
         message_format=args.format,
     )