Setup intel-dcb nodes as dedicated policy compute nodes

In order to accurately report quota use on Cheaha including files migrated to Ceph, we need to run a policy that will estimate KB_ALLOCATED for Ceph files as if they were on GPFS and include that in the quota report. This is a fairly simple process but requires aggregation across whole GPFS filesystems/filesets which can be large. Setting these to run on compute hardware used by researcher will be unreliable since node availability is variable based on job load. We don't want to run this solely on the ngenea nodes since those also have to perform transparent recall, and these policy runs should be run fairly regularly. Additionally, the ngenea nodes do not have Slurm installed and so can't submit jobs themselves.

Instead, we should use some or all of the intel-dcb nodes as reserved nodes for these runs so they can run on-demand. Not exactly sure what the technical requirements for this are so need to contact Datacore for their opinion. The PixStor Python API provides a defaultHelperNodes setting we could list these nodes under, but they can also be explicitly named in the API call.