node.fs.usedBytes reported by /api/v1/nodes/{NODE}/proxy/stats/summary doesn't match with the results of pod ephemeral-storage usage + node.runtime.imageFs.usedBytes

OCP 4.12

All the time

1. Get the amount of ephemeral-storage used reported by api: oc get --raw /api/v1/nodes/{NODE}/proxy/stats/summary | jq -r '.pods[] | ."ephemeral-storage".usedBytes' 

2. Get the amount of usedBytes for imageFS: oc get --raw /api/v1/nodes/{NODE}/proxy/stats/summary | jq -r '"imagefs:\(.node.runtime.imageFs.usedBytes)"'

3. Get the amount of usedBytes for nodeFS: oc get --raw /api/v1/nodes/{NODE}/proxy/stats/summary | jq -r '"Used:\(.node.fs.usedBytes)"'

4. ephemeral-storage.usedBytes + node.runtime.imageFs.usedBytes != .node.fs.usedBytes 

- The sum of reported sizes for usedBytes imagefs + the sum of ephemeral-storage for pods on the node doesn't match to what the node.fs.usedBytes reports

- Would like to understand how imagefs plays in node storage consumption.

- Reported stats should match to what is used within the node OR
- If the calculation is not correct, please share the details of how those stats are being calculated
- Need assistance from cAdvisor / kubelet engineer to clarify stats behavior

- Customer would like to understand:
  - How is this difference calculated?
  - What is the usage of metrics contribution to storage, node usage as customer wants to adjust some limits on ephemeral-storage
  - What is the pod ephemeral usage metrics such as imagefs , pod ephemeral usage etc and their contribution to the node disk usage?