chore: Performance optimization

[laszlocph]

Capacitor loads in 16 seconds in a cluster with 20 HelmRelease, 49 Kustomization and 150 deployed services. Capacitor loads in 110 seconds in a cluster with ~50 HelmRelease, ~300 Kustomization and ~500 deployed services.

Ideas:

• filter on the backend, make namespace selector mandatory
• load by data type: service list, pods, configmaps etc. With skeleton loader
• or both of the above

[laszlocph]

Reduced the 16 seconds to 3.5 the 110 seconds to 21

[hoerup]

Just noticed that the /api/services response includes .metadata.managedFields is that needed? if not, then by removing it you could save a huge chunk of transferred data ?