Catching Travis core dumps

[worktycho]

I am now apposed to this. I think we lose more by leaving the container based infrastructure, than we gain in having the occasional core dump.

[Seadragon91]

The best is then to wait until there is a fix or some way to get the core in containers. I closed the pull request.

[Seadragon91]

Found this nice stackoverflow answer.
gdb starts Cuberite and if Cuberite exits, gdb will run the commands bt, "info threads" and "thread apply all bt". At last it exits with the exit code of Cuberite.

Tested this idea in a commit and it works fine 

Travis (crash): All runs, clang debug
Travis(no crash): All runs, clang debug

If you like this idea, then I will create a pull request 


Funny side note, now no core dump required 

[xoft]

That looks great, by all means push it in