Connector can send to the server chunks of memory adjucent to the query

This can happen because internal helper function incorrectly detects UTF16 surrogate pairs. That may lead that it incorrectly count query length in bytes and thus uses additionally some rather small(normally) area of the memory past the end of the query as part of the query for conversion into connection charset. This happens if unicode query string contains 0xD8..0xDB value in low byte.