Remove redundant encode

Unclear why this was originally introduced, in 00a919e. The commit mentions Python 3.3 specifically. That version is long EOL. Now, the test suite still passes after removing the encode.

tldextract/tldextract.py

@@ -366,7 +366,7 @@ def _decode_punycode(label: str) -> str:
     looks_like_puny = lowered.startswith("xn--")
     if looks_like_puny:
         try:
-            return idna.decode(lowered.encode("ascii"))
+            return idna.decode(lowered)
         except (UnicodeError, IndexError):
             pass
     return lowered