prometheus-node-exporter: declare cpu/mem requests and limits

[consideRatio]

We've seen the node-exporter pod be evicted because it has no resource requests/limits when what needs to be evicted are other pods to resolve the situation. Let's not allow that to happen!

• This relates to Set limits & resource requests on all core pods #90, about setting requests and limits for support chart's pods.

CPU (leap, 30 last days)

Note that the use dropped at some point. That was me upgrading the GKE based k8s cluster in #2337.

Memory (leap, 30 last days)

Why I choose the requests and limits

1. We don't want this pod to starve other pods of CPU if it bursts, but it seems it can only burst to ~3m CPU. Setting a limit on 5m CPU seemed fine.
2. We want to ensure the typical operation is guaranteed CPU wise, which seem to be ~0-3m CPU. Setting a request of 5m CPU seemed fine.
3. We want requests/limits of memory to be the same to avoid getting evicted by memory, and crashing reliably if this is too low instead. I saw that it has peaked at ~21Mi, so I went for 30Mi to be safe, knowing there is a few hundred Mi of memory available still to not cause issues with planned requests/limits in New default machine types and profile list options - sharing nodes is great! #2121.

[pnasrat]

Minor changes for comments/docs

This generally makes sense though.

[consideRatio]

Thank you for reviewing @pnasrat! I realized that I had copy pasted incorrect prometheus queries so I drew the wrong conclusions. It turns out this pod is peaking at very low CPU (~3m) and memory (~21Mi), so I've made updates on queries, screenshots, and requests/limits etc.

[pnasrat]

lgtm

[consideRatio]

Thank you @pnasrat for reviewing!!

[github-actions]

🎉🎉🎉🎉

Monitor the deployment of the hubs here 👉 https://github.com/2i2c-org/infrastructure/actions/runs/4608120591