Add Organization type

[glrf]

The Organization type is provided by a custom API server. This requires the Organization resource to be in a different group than the CRDs. We introduce the group organization.appuio.io

Summary

This PR got quite large, however a lot of the code is auto-generated or tests.

I used the apiserver-runtime to reduce the amount of boilerplate, but the core business logic is independent from the library

• We introduce a new Organization type in apis/organization.
  • We use kubebuilder to auto-generate deepcopy but disable crd generation.
  • The type needs to be in its own group as we can only delegate whole groups to aggregate API server
  • This package also include functions to convert from and to namespaces
  • Most of the methods are there for apiserver-runtime to do its magic
• apiserver/organization contains the actual API server implementation
  • organization.go and organization_list.go contain the organizationStorage object that implements the rest.Storage interface of Kubernetes and is the core of the api server.
  • namespaces.go is only a thin layer of abstraction to access namespaces. I introduce it to more easily mock the interaction with Kubernetes in tests (See the auto-generate mock/ package)
  • table.go handles the server side printing for kubectl

I will handle deployment and adding the api server to the local-environment as a separate PR.

The CodeClimate issues are interfaces of Kubernetes

Checklist

• PR contains a single logical change (to build a better changelog).
• Categorize the PR by setting a good title and adding one of the labels:
  bug, enhancement, documentation, change, breaking, dependency
  as they show up in the changelog.

[simu]

LGTM overall, I looked at the following files carefully and didn't spot any errors:

• apis/organization/v1/organization_types.go
• apiserver/organization/organization.go
• apiserver/organization/organization_list.go
• apiserver/organization/table.go

Re codeclimate: Would it make sense to tune the maintainability checks to accept the methods which currently generate warnings (cf. https://docs.codeclimate.com/docs/advanced-configuration#section-default-checks)?

[ccremer]

Re codeclimate: You can manually make exceptions for individual cases if you're logged in (I did it for 2 of 3). They are persistent.
not sure if you want to address the max 4 allowed return statements (I assume it's classic Go error handling 🙄 ) in this PR or just ignore for now.

[glrf]

Re codeclimate: You can manually make exceptions for individual cases if you're logged in (I did it for 2 of 3). They are persistent. not sure if you want to address the max 4 allowed return statements (I assume it's classic Go error handling roll_eyes ) in this PR or just ignore for now.

It has nothing to do with Go's error handling. It's just a complex interface defined by Kubernetes that we need to implement and there is nothing we can do about it.

[ccremer]

It has nothing to do with Go's error handling. It's just a complex interface defined by Kubernetes that we need to implement and there is nothing we can do about it.

I'm pretty sure we could apply refactoring methods like "extract method" or introduce other helpers for it, but let's not make a fuss now ;)

[glrf]

It has nothing to do with Go's error handling. It's just a complex interface defined by Kubernetes that we need to implement and there is nothing we can do about it.

I'm pretty sure we could apply refactoring methods like "extract method" or introduce other helpers for it, but let's not make a fuss now ;)

Oh wait I completely misread the issue from codeclimate 🤦 I thought it complained about the number of return values..

So yes I guess we could "fix" that. But I'm not sure that would improve readability.

[bastjan]

LGTM, while I'm not super versed here I did not find any obvious code errors

🚀