Event generation around deployment states

[weyfonk]

Fleet already generates events related to git operations, such as received commits.
However, once git operations have succeeded, nothing is reported on the state of deployments.

Since the unit of deployment on a given target cluster is a bundle deployment, event generation could be added to the bundle deployment controller (part of upstream Fleet controllers), reporting at least failed deployments.

While this could lead to the Kubernetes API server being flooded with calls on setups involving large numbers of bundle deployments, client-go supports event aggregation with a configurable time window. That aggregation works against events with different messages, which would fit our use case.

Acceptance criteria

• Fleet generates events for failed bundle deployments
• A large number (e.g. 100 or 1000) of bundle deployments failing within a given window (e.g. 1 minute) lead to a single event being generated
  • Can that event contain information about every single failing bundle deployment?

This could be extended to successful bundle deployments too.