Document.get_type_name() returns a java.lang.String on arcadedb-jvm, a str on arcadedb #286

Document.get_type_name() (and so Vertex and Edge) returns a Python str on the native arcadedb build but a raw java.lang.String on arcadedb-jvm. graph.py returns self._java_document.getTypeName() unconverted on the JVM build, while the native wrapper returns the str it got in its snapshot.

The Java string compares equal to a str, so record.get_type_name() == "Person" works on both. It is not a str subclass, so string operations fail: "created " + record.get_type_name() raises TypeError: can only concatenate str (not "java.lang.String") to str on arcadedb-jvm and works on arcadedb. record.get(...), to_dict() values and get_rid() are Python str on both builds.

I found it while writing the functions and events docs page (#284), where a page example failed on the JVM build in CI. The page now documents the difference and tells readers to use str(record.get_type_name()), with a test that pins it.

The fix is one line in graph.py (str(...) around getTypeName()), plus a test that asserts type(...) is str on both builds. I can send the PR once #284 has merged, so the page loses that bullet in the same change.