Allow non ASCII chars as accelerator part 2

This is a follow up patch to the commit:
b80be502
This takes care of exposing non ascii characters to atk so that the
orca screen reader has a chance to present these characters to it's users.
I was considering removing the KeyCode bits all toghether and rely on
what we get from KeyChar but got cold feet in the end.

Change-Id: I7a3e78c47b81f8b36763c56e0298e8bbf226dccf
Reviewed-on: https://gerrit.libreoffice.org/19105Reviewed-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>
Tested-by: Samuel Mehrbrodt <Samuel.Mehrbrodt@cib.de>

vcl/unx/gtk/a11y/atkaction.cxx

@@ -203,6 +203,12 @@ appendKeyStrokes(OStringBuffer& rBuffer, const uno::Sequence< awt::KeyStroke >&
 
             if( c != '\0' )
                 rBuffer.append( c );
+            else
+            {
+                // The KeyCode approach did not work, probably a non ascii character
+                // let's hope that there is a character given in KeyChar.
+                rBuffer.append( OUStringToGChar( OUString( rKeyStrokes[i].KeyChar ) ) );
+            }
         }
     }
 }