Fix 156

[metacosm]

• Improve status computation so that it is more precise and can take different cases into account. Status reporting should also be improved.
• In particular, we are now able to change the status of the associated component when a link is created, using a newly introduced ComponentLinking status.
• This allows the controller to distinguish between a regular pending state and a state where the component needs to wait for its associated pod to be re-created: the component controller thus won't allow the component to switch back to ready until the associated pod has been recreated when the component is in the Linking state. The pod name will also be properly updated.
• This should properly fix #156 because it will cause hal to wait for the component to switch back to ready before attempting to push.

[cmoulliard]

When we execute this scenario using latest version of hal = 0.1.10 - https://gist.github.com/cmoulliard/9ea0351fd7848835f1b50335b3485e1f and that we check the link, we go bad status message

oc get links
NAME            AGE       STATUS    MESSAGE
backend-to-db   5m        Pending   Waiting for the following resources:
Component: Ready
client-to-backend   5m        Pending   Waiting for the following resources:
Component: Ready