PdfFontMetricsObject::UnicodeCharWidth returns width for CID instead of
unicode char (like does PdfFontMetricsFreetype and PdfFontMetricsBase14).
Maybe it can be fixed by converting unicode char into CID using
m_pEncoding->GetCIDValue (after few modifications). And implement also
Patch attached. I changes it to way as it is used in other places like
WidthExporter::update (which is seems used for font subset). In pdf
reference is stated that default width (DW key) is integer, then also
widths (W key) should be array of integers. But main purpose of this is
that object of