Update sonic.py

Sonic buffer accepts 20.000 bytes not unicode characters, since the chunking here is on unicode characters, sending 20.000 characters will overflow sonic's buffer.
UTF-8 can take up to 6 bytes, so sending less than (20.000 / 6) rounded minus should be ok.

archivebox/search/backends/sonic.py

@@ -5,7 +5,7 @@ from sonic import IngestClient, SearchClient
 from archivebox.util import enforce_types
 from archivebox.config import SEARCH_BACKEND_HOST_NAME, SEARCH_BACKEND_PORT, SEARCH_BACKEND_PASSWORD, SONIC_BUCKET, SONIC_COLLECTION
 
-MAX_SONIC_TEXT_LENGTH = 20000
+MAX_SONIC_TEXT_LENGTH = 2000
 
 @enforce_types
 def index(snapshot_id: str, texts: List[str]):