Skip cordon on nodes that are running some type of pods

[dbenque]

Today we can skip the drain activity on some kind of pods:

--evict-daemonset-pods     Evict pods that were created by an extant DaemonSet.
--evict-emptydir-pods      Evict pods with local storage, i.e. with emptyDir volumes.
--evict-unreplicated-pods  Evict pods that were not created by a replication controller.
--evict-statefulset-pods Evict pods that were created by an extant StatefulSet.
--protected-pod-annotation=KEY[=VALUE] ...

The problem is that even if we block the drain, the associated node is cordon. On a long run that could lead to have a lot of cordon nodes, leading to some operational issues.

Would you agree to add some options to exclude from cordon nodes based on the type of pods that they are running? Example:

--no-cordon-node-with-statefulset-pod
--no-cordon-node-with-unreplicated-pod
--no-cordon-node-with-protected-pod
--no-cordon-node-if-pod-cannot-be-drain  // this one is kind of a superset of all the other (but daemonset of course)

WDYT?

On the implementation side that would lead to run a pod informer, that would result in more resource consumption. More options, more power, but also more resources required :)

[dbenque]

I have some code ready for this, but it was built on top of a work that already integrate #99 and #90 , so waiting for these 2 PRs to progress before pushing a PR with the code:
https://github.com/DataDog/draino/commit/256dbecf37e1afd2e53c65090aeb0ae8be163001