chore: bump metrics and node pool (#10745)

We didn't have persistence enabled in our prod prometheus instances. 

This does that, and scales up the loki backend,

Also, increases the max size of the primary node pool to 256 since we
were close to the limit on several occasions.

Last, update the dashboard to reference namespaces by default.