Keep track of job duration in results

The time it takes for a job to run should be stored as a test result to be able to use it as some form of performance benchmark.

For example, when waiting for devices on the deferred probes list, the duration of the job can vary if devices take longer to probe. If they still probe within the time limit then it's not a failure, but nothing will be detecting the slowdown without a time measurement.

No files, tests, or entry points are named. Start by tracing where jobs run and where test results are recorded, then identify how deferred-probe jobs are represented. Done means job duration is stored with the result and can expose slowdowns without treating jobs within the time limit as failures.

Written by the indexing model from the issue text.