The file in question uses an undocumented Encoding type for the Arabic glyphs. The encoding format is as follows:
9 0 obj
[ 0 /uni060C
243 /uniFEFF ]
There is no mentioned of character names /uniHexCode in the PDF specification. If I were to guess this is a just format used by SAP. I've looked up the mappings 164->/uniFEB0 and the mapping appears to be valid, and corresponds to the character 'ARABIC LETTER ZAIN FINAL FORM' I need to figure out a way to detect and assign the correct glyph for this undocumented format.