This patch adds a manager argument that can be used to restrict the
resources being reconciled by the manager to a particular namespace.
This option can be used independently or together with the already
present watch-filter label to fine tune the number of resources being
targeted by the operator deployment.

This patch updates the watches of each controller to trigger a
reconciliation of a resource once a resource linked in a
ProviderConfigRef gets updated.

While a shorter requeue-interval is useful for local development and
tiny fabrics to quickly observe and correct configuration drift it's not
appropriate for production environments where a large fleet of devices
are already in a desired state but just require a constant sync loop
every hour. Therefore the defaults are adjusted. Users of the
network-operator can still configure a custom value through specifying
the respective container arg in their helm deployment to fine tune the
requeue-interval to their respective needs.

Add watches on Device resources to all controllers that use
annotations.IsPaused to trigger reconciliation when a Device's Paused
spec field changes. This ensures resources are re-reconciled when their
parent Device is paused or unpaused.

Each controller now:
- Watches Device create/delete events (default behavior)
- Watches Device update events only when Spec.Paused changes
- Ignores Generic events for Device watches
- Uses deviceTo* map functions to find associated resources by label