Resource view - reconcile inconsistencies between metrics of different topologies

[fbarl]

As @2opremio has predicted, having multiple sources of truths for the nodes' metrics between hosts, containers and processes topologies is making some processes in the resource view seem as if they were consuming more resources than their parent containers.

To prevent resource boxes rendering overflows (and overlaps) in such cases, a temporary hack is used that renormalizes the total length of the overflown children to perfectly fit on top of their parent. This is just a visual fix though and it doesn't affect the displayed metric info in any way.

A proper solution would probably require backend to draw all the metrics info from a single source so that:

• the total resource utilization of children could never exceed the parent's
• resource capacities info was (more) consistent between different topologies

[rade]

The underlying problem is #1133.