Job view for single node GPU jobs with only node level metrics fail to render

[moebiusband73]

The job view for single node (and probably also) shared GPU accelerated jobs fail to render properly.

[spacehamster87]

Frontend will check if all requested accelerator metrics are scope:accelerator in the cluster configuration before fetching:

• If true, additionally fetch accelerator scope by default
• If one metric does mismatch, only load node, (core)

In the latter case, if metrics do have accelerator scope, it can still be fetched by using the load more... optin in the plot-scope dropdown.