LibPDF: Fix handling of differences array in custom encodings

When looking up differences in the specified encoding, we previously
didn't recognize a lot of characters, namely those that are referred to
by a string in the PDF itself, like "/germandbls".

We now create a mapping of those characters to the code points they are
referring to, and correctly look them up when needed.

Userland/Libraries/LibPDF/Encoding.h

@@ -647,6 +647,7 @@ public:
 
 protected:
     HashMap<u16, CharDescriptor> m_descriptors;
+    HashMap<String, u16> m_name_mapping;
 };
 
 }