Description:
------------
currently set_exception_handler() returns null on error, and null|callable on success. 

so when it does return null, there's no good way to tell "did it return null because of an error, or did it return null because of success-but-there-was-no-previous-handler ?"

.. i think it would be an improvement if it returned bool(false) on error and null|callable on success, then it would cleanly separate error-returns from success-returns. i guess that's a BC-break and should go in the next major version though (for scripts knowing full well that there's a previous handler and looking for nulls to detect errors? idk if that even exist, but i guess it might)