We are spending a lot of time computing issue hashes

[marco-c]

E.g. in https://firefoxci.taskcluster-artifacts.net/ErAfZ8x-Tymdm4O5QCsU8A/0/public/logs/live_backing.log it's more than a minute (where the total runtime of the task is 3 minutes and 30 seconds.

We could calculate issue hashes in parallel instead of sequentially.