Thread pool fails to finish gracefully when a thread is dying

[mkuemmel]

Looks like when there is an exception in a thread, SE++ does not know what to do. The SE++ just hangs around.

That behavior showed up in #442 and #438 and camouflaged the original problem there.