[PATCH] malloc: Abort on heap corruption, without a backtrace [BZ #21754]

[Florian Weimer]

I'm going to push this with the NEWS edit suggested by Carlos, and with
the error message change reverted, per Paul's recommendation.

I still think we need to adopt a more consistent formatting for
process-terminating error messages and document them in the manual, but
this change is not the place for this discussion.

Thanks,
Florian