* Based on recent discussions in #68 we think rescuing all
exceptions like this was probably a mistake.
If this continues to be a problem we will need to narrow things
down to a reproducible issue we can report to the RSpec team.
One possible work-around might be to catch exceptions that occur
within the specs themselves, and to bubble up everything else.
However, this does bring up the question of why the specs failed;
if it was due to a syntax error caused by mutant and unparser,
then that's probably a bug we need to fix. It could be for some
other reason, which we would have to isolate and handle
explicitly, but I think we are all in agreement that this logic is
too broad.
59f952d56f