Define default global Kamelets

[nicolaferraro]

We should define a concept of global Kamelet, i.e. a general Kamelet that is not local to a specific namespace, but can be used from any namespace. By default (in the PR #1602), integrations can only use Kamelets contained in the same namespace.

Having global Kamelets can allow UI's built on Camel K to show "Sources" or "Sinks" available in the cluster that are ready for use.

When installing the operator, we can install by default some global Kamelets (defined in this repo or, in the future, in a "Kamelet hub") in a special namespace, so that graphical tools can pick them up.

The issue is where to put them:

1. A system namespace (like kube-system, openshift)? Seems not a standard approach
2. A camel-system namespace? Operator needs permissions to create that namespace and read kamelets from it (cluster-role)

Then we should:

• Adapt the operator runtime to lookup that namespace and allow to use it: resolution will be first attempted on the local namespace, then on the global one
• Install some global Kamelets automatically on installation (and keep them updated on new releases): initial set of kamelets can be picked from the examples provided in the https://github.com/apache/camel-k-examples repository for Knative sources