fix(coordinator): reset roller state when proof session timeout

[colinlyguo]

Purpose or design rationale of this PR

Symptom:

1. Coordinator offline.
2. Roller submits and fails.
3. Coordinator runs again.
4. Roller connects again.
5. Proof session timeouts (because the task is popped and discarded, even if the submission fails).
6. The coordinator won't assign a new task to the roller.

Analysis:
In the above scenario, the coordinator does not run freeTaskIDForRoller since the roller discards the task itself and never submits the proof. Then the coordinator would always treat the roller as not idle (won't assign a new task to the roller).

Fix:
Run freeTaskIDForRoller when proof session timeouts.

Does this PR involve a new deployment, and involve a new git tag & docker image tag? If so, has tag in common/version.go been updated?

Yes, and updated.

Is this PR a breaking change? If so, have it been attached a breaking-change label?

No

---

• To see the specific tasks where the Asana app for GitHub is being used, see below:
  • https://app.asana.com/0/0/1203663360337455

[0xmountaintop]

let also bump the version number

[colinlyguo]

The reset may cause the coordinator to assign tasks to slow provers, thus the tasks assigned to these slow provers may be delayed.

[colinlyguo]

TODO: this is a temporary fix, will replace coordinator reset roller state when timeout with roller reports healthy status.

[ChuhanJin]

[ChuhanJin]

[ChuhanJin]

[ChuhanJin]

[ChuhanJin]

[ChuhanJin]

[ChuhanJin]