fix: deduplicate jobs from getJobsForWorker to prevent parallel execution of same job

[oharbo]

Problem

When a worker has concurrency > 1 and there is only one job in the queue, the native getJobsForWorker(name, count) can return the same job multiple times (to fill the requested count). The JS layer then runs that job in parallel on multiple workers, causing:

• One execution may fail (e.g. file not found, compression error)
• Another may succeed
• Inconsistent state: same task id appears as both FAILED/REMOVED and UPLOAD OK
• onQueueFinish receives duplicate entries in executedJobs

Solution

Deduplicate jobs by id before execution. Keep only the first occurrence of each job id.

Reproduction

1. Register a worker with concurrency: 2
2. Add a single job to the queue
3. Observe the job being executed twice in parallel (e.g. via logs or onQueueFinish with 2 entries for the same job id)