Should all idempotent! jobs be deduplicated
In #42 (closed) we've started deduplicating all idempotent jobs using the most restrictive until_executing
strategy. This means that we'll drop the second job when it is enqueued while there is already a job in the queue, for the same worker and arguments.
Is this over-eager? Should this be explicitly specified on the job instead?
This loosely relates to #195 (closed), where we would allow developers to specify different deduplication strategies. Perhaps we should implement a dont_deduplicate
strategy that developers can explicitly set if they encounter a job that should not be deduplicated?