When we send a command through the NetworkInterfaceTL, we schedule a killOperations command if a failure occurs and an operation key exists on the request. This killOperations command has historically caused test failures that were not expecting an extra command to run. We should update NetworkInterfaceTest::makeTestCommand() to not create a request with an operationKey if one is not passed in. That way, only the tests that expect a killOperations will schedule one.