fix: clear _continuation_expected before decode_body in failure branch

Moved the flag clear to before FailureResponse.decode_body() in
decode_continuation(). Previously if decode_body itself raised (e.g.,
truncated failure body), the flag stayed True alongside a poisoned
buffer. Now the flag is always cleared when the continuation sequence
is terminated, regardless of whether subsequent operations raise.

Closes #125

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: antoineleclair

src/dqlitewire/codec.py

@@ -302,8 +302,8 @@ def decode_continuation(self) -> RowsResponse | None:
                 )
 
             if header.msg_type == ResponseType.FAILURE:
-                failure = FailureResponse.decode_body(body, schema=header.schema)
                 self._continuation_expected = False
+                failure = FailureResponse.decode_body(body, schema=header.schema)
                 raise ProtocolError(
                     f"Server error during ROWS continuation: [{failure.code}] {failure.message}"
                 )