Proper Node Metrics expose for DB instances created with DBaaS

Impact on the user:

• The user is unable to see the proper "node/container" parameters of OS-level metrics in DBaaS

STR:

1. Setup PMM 
2. configure DBaaS
3. provide correct  Server Address in settings
4. Start default small MySQL DB Cluster  (3 nodes, 1 cpu, 2 GM memory, 25 GB)
5. Go to dashboards, Mysql Instance summary 

Expected Result:

• At the bottom, the user can see Node/container metrics 
  memory 2 GB, 25Gb disk

Given Result:

• some strange/big numbers. Probably all cluster parameters, not specific Mysql
  • One can see memory available on kubernetes node, not the container itself because /proc/meminfo is not namespaced. Thus even though container is limited by cgroups to 2GB of memory, it could not be seen.

jan@jan-Latitude-5591:~$ kubectl exec -it df7143-pxc-0 -- bash
Defaulting container name to pmm-client.
Use 'kubectl describe pod/df7143-pxc-0 -n default' to see all of the containers in this pod.
bash-4.2$ free
              total        used        free      shared  buff/cache   available
Mem:       16417904     3957868    10273744      642392     2186292    11869520
Swap:             0           0           0

Workaround:
N/A

 Suggested implementation

• use https://github.com/google/cadvisor instead of node_exporter

Original report: 

Notes:
https://github.com/google/cadvisor/blob/master/docs/storage/prometheus.md