In order to streamline the process for debugging CQ issues, we should have more failure details added to the messages it sends. Those details should include:

1. The names of the OST suits that failed

2. The OST tests that failed

3. If it was a build failure rather then an OST failure

4. If it was possible to detect - the particular change(s) that caused failure.
   #if its possible to tell - if its an infra issue or a code regression

To implement this we would need at the very least a richer interface for passing information between the CQ tester and the and manager jobs - possibly some kind of a structured data file passed around. We may also need a richer interface to OST itself (to tell which test failed) and perhaps even to the build jobs (to tell why the build failed).