[Core] Scheduler policy for adaptor failures

[luis100]

Define policy from pronom and c3po failures, and based on:

• Where failures come from: known (PluginException), unknown (Trowable), system (jena, etc.)
• In which state should the tasks be restarted
• How many times should they restart (e.g. 3 with the same state, 3 with the next batch, 3 with reset state?)
• Should the plugin make the decision of where to start when failure occurs?

[luis100]

• Create a FatalPluginException and TorelablePluginException. If it torelable then continue with the next batch.
• When restarting the plugin, do not start immediatly but wait for the next schedule (period)