Coordinate Spin kube plugin and Spin Operator SpinApp CRD versions

[kate-goldenring]

There is currently no clear way to know which version of the spin-kube plugin is for which version of the Spin Operator. This is important to ensure they both are referencing the same version CRDs. Some options I see are

1. Clearly document a compatibility matrix
2. Have the plugin take in a Spin Operator version parameter to determine which CRD to use
3. Match versions on releases (aka spin-kube v0.3 is compatible with Spin Operator v0.3)

My preference would be option 2 with a default to latest

A motivating example for this is PR #87 which uses a newly added field to the SpinApp CRD

[bacongobbler]

For spin kube scaffold, it only scaffolds out some text files that can be applied to Kubernetes. It has no knowledge about what's installed in your cluster, including what version of SpinKube is running.

The only commands that have some knowledge of your kubernetes cluster's state is spin kube deploy, spin kube connect, spin kube delete, etc. All of which are hidden commands - we might not support those commands moving forward.

If we are to remove those commands and just scaffold out files, my preference would be to write out CRDs based on the latest release of SpinKube. If users want to scaffold files for prior versions of SpinKube, I'd prefer documenting a compatibility matrix and a set of instructions on how to install those versions (via spin plugin install --version).

If we are to make spin kube scaffold a little smarter, I'd recommend a fourth approach: looking up what schemas are installed in the kube cluster and scaffold out whichever is latest, falling back to option 2 in the event of a cluster not being present. I'm assuming this is something clients can do via the kubectl get crds APIs (or whichever API endpoint is most appropriate for this).

All that being said, I'm okay with punting this down the road while we're still in v1alpha1, Once we start moving to v1beta1 and we plan to support both APIs, then we might want to perform some form of schema validation against the server. Do we have an idea when a v1beta1release of the schema might come out?

Thoughts on all this?