Add a jstestfuzz_interrupt_sharding suite

We will be allowing users of Atlas to have the killOp and killSessions* privileges. It is important that no matter what operations they kill, including internal intra-cluster or intra-node operations, that the cluster continues operating in a healthy way. We have jstestfuzz interrupt suites for standalones and replica sets, but not for sharding. We should make sure that all sharding internal operations are safely interruptible.