Checking status of re-run jobs overestimates progress

1. Project is run through pipeline. Status checks work fine.
2. Project completes, but an insufficient proportion of reads were merged so the job is restarted as single-end.
3. The status-checking process looks for the presence of log files for steps that ran in step 1, rather than in step 2.

Most straightforward fix would be to move the `slurm_logs` directory somewhere else, we just have to remember to add it to the archive if the project eventually succeeds.