I stumbled upon an instance triggering a segfault with a v1.14 release build. Using a debug build hits a capacity-related assert.

The instance includes a mix of jobs and shipments along with various priority levels and it turns out the problem is caused by an invalid application of the PriorityReplace local search operator introduced as part of #1021. The operator splits a valid route at a point where there are still pending deliveries so we end up in an invalid state.