Investigate the use case of slurm sacct

[khsrali]

It seems aiida-core is using this to get the exit code in a complicated manner:

tasks.py and calcjob.py both expect a dictionary with three keys ('retval', 'stdout', 'stderr') from:
scheduler.get_detailed_job_info()
Which then calcjob.py uses it along with two other files to call again on scheduler.parse_output to get the exit code.

[khsrali]

Without this feature, monitoring cannot be done.
For now, in case of a failure, not possible to know which of these caused it:
ERROR_SCHEDULER_OUT_OF_MEMORY
ERROR_SCHEDULER_OUT_OF_WALLTIME
ERROR_SCHEDULER_NODE_FAILURE etc..