Is your feature request related to a problem? Please describe.
The ILP scheduler gives up after 10 seconds, falling back to the greedy scheduler. ILP is important for massive jobs where greedy (and presumably) breadth-first execution of the DAG would use up more disk space than is available, but disk-intensive operations also tend to saturate system I/O such that ILP takes way longer than 10 seconds.

Describe the solution you'd like
A command-line option, also set-able via config.yaml, to customize the ILP scheduler timeout duration.

Describe alternatives you've considered
The best alternative I've found thus far is to SIGTERM and restart the workflow every once in a while.

Additional context
Thanks for the banger software btw