Reimplement the feature of resubmitting failed index tasks

[taojing2002]

Currently the class of IndexGeneratorTimerTask.java in metacat-index can resubmit the failed index tasks periodically. However, the implementation directly uses the hazelcast code, which is obsoleted in this release. We need to reimplement the class in the metacat component.

[taojing2002]

We found there is a database table (index_event) and java class IndexEventDAO behind hazelcast. So now we directly use the code in Metacat.

[mbjones]

@jeanetteclark could you please review this with @taojing2002 to ensure the rabbitmq tasks are fully handled in Metacat, based on your experience implementing it for MetaDIG?

[taojing2002]

Now, the feature, which resubmits the index tasks which fails to be put into RabbitMQ (e.g., the RabbitMQ is down), has been done.

However, we still need to think about how to resubmit the index tasks, which fail in Dataone-Index.

[artntek]

see DataONEorg/dataone-indexer#23