I'll take a wild guess:
The file contains unicode sequences for extended chars and the diagnose is based on (mismatched) length.
if I debug the code in the it looks like "<\0h\0t\0m\0l\0>\0<\0h\0e\0a\0d\0>\0\r\0\n\0<\0M\0E\0T\0A\0 \0h\0t\0t\0p\0-\0e\0q\0u\0i\0v\0=\0\"\0C\0o\0n\0t\0e\0n
Which is a valid beginning of a HTML file except for the very first char. The file is probably damaged by missing a unicode marker at the start. This damage was probably caused when it was written and is not (easy) repairable now.
You could try setting the WebClient.Encoding to UTF8 (and try a few ASCII as well).