Need a way to clean up orphaned runs

## Background
Currently, if a job is interrupted in a way that it didn't have a chance to send the failure request, the status of the job is 'Running' forever.

## Feature
We need to periodically check all jobs that are in 'Running' state for days and switch them to 'Failed' with the appropriate error.

## Proposed Solution [Optional]
1. Such periodic cleanup can be done in Menas itself. Checking interval can be 1 hour by default and can be configurable.
2. The maximum age of a job can be a configuration parameter as well.