Parallel cluster configuration cannot be changed if clusters don't initialize.

[mducle]

The parallel cluster stuff does not work for pace-python at the moment. However, the configuration insists on checking its instance first before allowing you to change the configuration. The problematic code is check_and_set_cluster_.m:11. This seems to require the initialization of the current configuration even if it is the "automagic" setting which may not work.

So on pace_python running in IDAaaS, the system detects that it should be running a cluster. When running:

hc = m.parallel_config()
hc.parallel_cluster = 'none'

This error is given:

Error using ClusterSlurm/init_queue_parser (line 342)
Invalid default value for property 'known_clusters_' in class 'MPI_clusters_factory':
 Can not retrieve user name. Error: /bin/bash: whoami: command not found

MatlabRuntimeError: An error occurred when evaluating the result from a function. Details:
  File ~/.mcrCache9.8/pace_50/pace/CTF/Herbert/classes/MPIFramework/@ClusterSlurm/ClusterSlurm.m, line 342, in ClusterSlurm.init_queue_parser
  File ~/.mcrCache9.8/pace_50/pace/CTF/Herbert/classes/MPIFramework/@ClusterSlurm/ClusterSlurm.m, line 107, in ClusterSlurm.ClusterSlurm
  File ~/.mcrCache9.8/pace_50/pace/CTF/Herbert/classes/MPIFramework/@parallel_config/private/check_and_set_cluster_.m, line 11, in check_and_set_cluster_
  File ~/.mcrCache9.8/pace_50/pace/CTF/Herbert/classes/MPIFramework/@parallel_config/parallel_config.m, line 331, in parallel_config.set.parallel_cluster
  File ~/.mcrCache9.8/pace_50/tmp/pip-req-build-c5c20nv1/src/call.m, line 69, in callInvalid default value for property 'known_clusters_' in class 'MPI_clusters_factory':
 Can not retrieve user name. Error: /bin/bash: whoami: command not found

[oerc0122]

Yes, part of the MPI_clusters_factory empty clusters in a known_clusters_ map, which should probably be changed to function references?