Job Failure Messaging

[dfaller]

Transition job execution failure in the scheduler to the backend messaging system.

1. Create command message for updating recipe dependent job statuses
2. Create command message for failing jobs
3. Change scheduler task update handling to use failing jobs message
4. Change task handling thread to use failing jobs message for timed out jobs
5. Change scheduler lost node callback to use failing jobs message
6. Change scheduler thread to use failing jobs message for executions that are starved or have a lost node
7. Remove all error model retrieval in the scheduler?