Checking status of re-run jobs overestimates progress

[rabdill]

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.