Add snapshot validation webhook

[timoreimann]

Add the snapshot validation webhook responsible for labeling invalid snapshots. This is the recommended approach in preparation to installing snapshots v1.

See the upstream docs for details.

[varshavaradarajan]

@timoreimann - would this cause problems in legacy master droplet not being ready after an upgrade?

[timoreimann]

@varshavaradarajan if the webhook wasn't ready or working, then snapshot creates and updates would fail. Other operations should not be affected. In particular, it should not affect the general provisioning procedure.

Not 100% sure if it answers your question. If not, please let me know.

[varshavaradarajan]

There have been some support issues where user configured webhooks with FailurePolicy Fail has caused problems right after an upgrade making the master droplet unready. Do we need some sort of check in the bootstrap-master script where initially, the FailurePolicy of this webhook will be Ignore and then after the master is ready, we can set it to Fail?

[timoreimann]

@varshavaradarajan I think the kind of problems you are referring to are about failing web hooks that prevent critical components (e.g., Cilium) from coming up after an upgrade. In contrast, the snapshot validation webhook only applies to volume snapshot create and update requests, so it should not be a problem. At least, I cannot think of a scenario where a failing snapshot request could block provisioning, especially since the CSI driver should retry and eventually succeed once the web hook is up and running.

In the end though, it's up to the concrete user of our FOSS CSI driver to configure the web hook to their needs. As the changes to the README explain, the setup of the webhook in this PR is mostly to provide a (non-production-ready) example (and to enable easy end-to-end testing integration).