Document use cases and expectations

[nikos912000]

The chaos controller supports a wide range of failures at different levels and on different targets.

I think it would be incredibly useful to provide use cases and expected behaviors.

Common questions in chaos engineering:

• "Why would I run this experiment. What is the use case?"
• "What should I expect from this experiment"? For certain experiments the bevahior is clear; e.g. terminating a Node. For others not completely. E.g. a kernel panic wouldn't trigger the autoscaler which replaces the Node, but the workloads running on that Node will get affected.

Internally we are documenting past experiments teams have run over time. This helps our users but is tailored to our platform and needs.

What do you think of introducing docs for that? We can both make our use cases generic enough to fit into this repository.