We have a number of tests which output words "failed" and "exception" when the failures and exceptions are expected. This adds significant cognitive load when analyzing the logs.

Ideally, the tests should produce the output only when they fail.
Also, the tests should fail when unexpected exception is thrown, especially the headful ones.

Solution
======
Add methods to ErrorLoggingUtility which
- redirect stderr to an in-memory buffer
- allow for checking thrown exceptions in the buffered output