find a way to add a timeout to dot, or kill it when the current process is killed
When we try to render a very large automaton with dot, it sometimes takes years. If we are in a jupyter notebook, killing the python kernel do not seem to kill the dot
process. That is an issue on our spot-sandbox
VM.
Can we arrange things so that dot is automatically killed when the parent Python process is killed? Or should we run it with a timeout? (I'm a bit afraid that forcing timeout my create some spurious failure in the test suite on overloaded hosts.)