Interrupt in NTL malloc() hangs Sage

[jdemeyer]

Inspired by #16667, the following usually breaks Sage badly:

sage: x = polygen(GF(2^8, 'a')); alarm(0.1); x^1000000

This reason is that this tends to interrupt during a malloc() call, which can totally mess up the heap. We always knew this could happen, but now we have a case which is relatively easy to reproduce.

CC: @vbraun @jpflori @defeo @pjbruin @sagetrac-sbesnier @nexttime

Component: cython

Keywords: interrupt

Issue created by migration from https://trac.sagemath.org/ticket/16850