BOM (Byte Order Mark) in the HTML file changed the underlying eSrcEnc
encoding, but did not actually update the rtl_TextToUnicodeConverter hConv.

Subsequent changes of eSrcEnc in SetSrcEncoding() (triggered by
'content="application/xhtml+xml; charset=UTF-8"' in the HTML file) were then
ignored (eSrcEnc was already set to UTF-8), and the parser was happily using the
old (Windows-1250) hConv.

Change-Id: If432d59891d51c6abe3517e325ed73057d0f8610