DistributedEngine silently ignores when grid-jobs are canceled (because of exceptions) on grid-clients

[gkronber]

When an engine is cancled at a grid-client because of an exception it will silently ignore the error and tries to execute the operation again locally. This is not good because it could go important errors undetected.

Instead it would be better to open a message-box showing the exception which was thrown on the grid-client (for each canceled job) and abort the distributed engine (requeueing a combined-operation containing the operations that were cancled (optionally merging the results of the successful operations). This way the error-handling of the DistributedEngine would be similar to the behaviour of the SequentialEngine or ThreadParallelEngine.