From feedback collected via NETOBSERV-1005
The dashboards should present metrics filtered for infra and/or for application, similar to what we have in the plugin.
We don't need to change the way metrics are generated, it can be done from the queries, e.g:

• either by introducing a custom toolbar filter infra/apps (but I'm not sure if the console dashboards support that)
• or by splitting each chart in two: one filtered for apps, the other filtered for infra

in both cases, filtering can be achieved with promql label selectors and unions. Examples:

// Infra flows:
topk(5,sum(rate(netobserv_namespace_flows_total{SrcK8S_Namespace=~"openshift.*"}[1m]) or rate(netobserv_namespace_flows_total{SrcK8S_Namespace!~"openshift.*",DstK8S_Namespace=~"openshift.*"}[1m])) by (SrcK8S_Namespace, DstK8S_Namespace))

// Non-infra flows:
topk(5,sum(rate(netobserv_namespace_flows_total{SrcK8S_Namespace!~"|openshift.*"}[1m]) or rate(netobserv_namespace_flows_total{SrcK8S_Namespace=~"openshift.*",DstK8S_Namespace!~"|openshift.*"}[1m])) by (SrcK8S_Namespace, DstK8S_Namespace))

IMPORTANT: as a pre-requisite we should probably move away from using FLP confgenerator for generating this dashboard, as it is limiting our ability to tune the dashboard as desired. For instance we'd like to provide the right unit for Y axis, or to display charts side by side rather than taking full rows. This would make the story a bit bigger for estimates.