prune Cloudsql

[ibawt]

@KnVerey is there anything that's going to blow up if we remove cloudsql from the prune list? We've had a few issues due to dangling resources.

[KnVerey]

That isn't easy to do because

• the whitelist would need to become dynamic based on whether or not cloudsql is actually available in the target cluster (this is part of the dynamic resource discovery work @stefanmb was doing in Runtime resource discovery #188); otherwise, the tool will cease to work against clusters that don't have that resource.
• custom resources are currently deployed with create/replace, not apply. When we upgraded to 1.8, we ran into errors related to the fact that custom resources don't support strategic merge patching (and according to the documentation, this is still true). We don't currently support pruning anything that uses the replace strategy. It might work to batch them into the mass-apply after they've actually already been deployed, but it might not. Related: Use our own annotation to drive pruning #254

[ibawt]

blech.

[ibawt]

cc @mkobetic

[dturn]

as an update #312 has been merged. So the first bullet has been accomplished

[dturn]

The kubernetes-deploy side of this work has shipped.