Extend `-trace` option with a coredump
On must linux machines programs will cause a coredump if there is a memory problem. Our
judft_error call don't do this. It would be nice if we could somehow trigger this. (I have seen some examples for C/C++). An option similar to
-trace would be nice.
These core dumps can be loaded into a debugger (gdb or totalview) and then one can see the whole program state rather than just the position.