Standardize on nodepool labeling conventions

[yuvipanda]

From @consideRatio in #134 (comment):

The configuration scheduling.corePods.nodeAffinity.matchNodePurpose=require will make all core pods of the jupyterhub chart have a required node affinity toward the node label hub.jupyter.org/node-purpose=core.

By using hub.jupyter.org/pool-name it seems like there is a predefined meaning of that label and can cause some confusion, it isn't having a predefined meaning right? If it is a label with meaning for the 2i2c.org pilot hubs, I think we should name it like 2i2c.org/pool-name instead.

We should investigate either node-purpose or switch to a different namespace.

[consideRatio]

By using hub.jupyter.org/pool-name it seems like there is a predefined meaning of that label and can cause some confusion, it isn't having a predefined meaning right? If it is a label with meaning for the 2i2c.org pilot hubs, I think we should name it like 2i2c.org/pool-name instead.

The k8s label hub.jupyter.org/pool-name is no longer in use, we are instead referencing the k8s label node.kubernetes.io/instance-type for such purposes I think. We have also stopped using the hub.jupyter.org label prefix for labels we define (#1256).

I'll go for a close here, lets open a new issue if there is some concrete aspect to focus on remaining that I've failed to capture.