chore: Small style + usability refactors for new contributors

[rakechill]

Description
These changes are based on feedback from @matthchr

• Update (and as a follow-up git ignore) skaffold.yaml so that make az-all automated local changes during development are less confusing.
• Style refactors
  • NewAZClientFromAPI vs. NewAZClient. Maybe one of these should call the other then? NewAZClient constructs all the clients + calls NewAZClientFromAPI
  • Inconsistent with abbreviation casing (CreateAzClient vs. NewAZClient)
  • Fakes: var _ instance.NetworkInterfacesAPI = (*NetworkInterfacesAPI)(nil) ->
    var _ instance.NetworkInterfacesAPI = &NetworkInterfacesAPI{}

How was this change tested?

• make presubmit
• make az-all

Does this change impact docs?

• Yes, PR includes docs updates
• Yes, issue opened: #
• No

[Bryce-Soghigian]

@rakechill there is more than a single test using normal go tests.

./pkg/auth/cred_test.go
./pkg/auth/config_test.go
./pkg/providers/imagefamily/bootstrap/aksbootstrap_test.go
./pkg/fake/skus_test.go
./pkg/fake/azureresourcegraphapi_test.go
./pkg/utils/gpu_test.go

Are all the go tests. IMO sometimes its nice to write a very simple go test rather than setting up all the ginkgo shenanigans and they are better for fine grained unit tests vs the coarse unit tests we write through the fakes framework.

[Bryce-Soghigian]

networkinterfaceapi.go fake BeginDelete incorrectly returns a nil poller

this is addressed in my pr for networkinterface delete

[rakechill]

@rakechill there is more than a single test using normal go tests.

./pkg/auth/cred_test.go ./pkg/auth/config_test.go ./pkg/providers/imagefamily/bootstrap/aksbootstrap_test.go ./pkg/fake/skus_test.go ./pkg/fake/azureresourcegraphapi_test.go ./pkg/utils/gpu_test.go

Are all the go tests. IMO sometimes its nice to write a very simple go test rather than setting up all the ginkgo shenanigans and they are better for fine grained unit tests vs the coarse unit tests we write through the fakes framework.

@Bryce-Soghigian Ack. Maybe a better solution is to call out the different sets of testing we do instead. That could go somewhere on the contributing.md/website. I've written something up in my notes in the meantime. Happy to port over/create a PR as desired.