fix: avoid wrapping RawError twice

[marwanad]

What type of PR is this?

Uncomment only one /kind <> line, hit enter to put that in a new line, and remove leading whitespaces from that line:

/kind api-change
/kind bug
/kind cleanup
/kind design
/kind documentation
/kind failing-test
/kind feature
/kind flake

What this PR does / why we need it:
The sender decorator builds a local Error type that embeds the RawError returned from the Azure service. However, in the client's sendRequest method it wraps this error again under the Error type.

retry.GetError(response, err)

This would return a double-wrapped error the consumers:

Retriable: false, RetryAfter: 0s, HTTPStatusCode: 409, RawError: Retriable: false, RetryAfter: 0s, HTTPStatusCode: 409, RawError: {"error":{"code":"ReadOnlyDisabledSubscription","message":"The subscription 'xxx' is disabled and therefore marked as read only. You cannot perform any write actions on this subscription until it is re-enabled."}}
 

Initially was thinking of safe guarding the behaviour but because Error doesn't implement the error interface, it seemed to be a bit of hassle.

With the fix in place, the top-level consumers will get the expected error body:

Retriable: false, RetryAfter: 0s, HTTPStatusCode: 409, RawError: {​​​​​​​"error":{​​​​​​​"code":"ReadOnlyDisabledSubscription","message":"The subscription 'xxx' is disabled and therefore marked as read only. You cannot perform any write actions on this subscription until it is re-enabled."}​​​​​​​}​​​​​​​

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Release note:

N/A

[marwanad]

@feiskyer any reason we don't embed the ServiceError type instead of RawError? It seems that every code-path should return this.
https://github.com/Azure/go-autorest/blob/82fc42f340cec32d30ce97f9277f9b8aaf8885df/autorest/azure/azure.go#L50-L57

[coveralls]

Coverage decreased (-0.04%) to 80.695% when pulling 7d447e2 on marwanad:avoid-emebdding-raw-error-twice into cb84088 on kubernetes-sigs:master.

[feiskyer]

looks good

/retest

[feiskyer]

any reason we don't embed the ServiceError type instead of RawError?

ideally, every azure api should report ServiceError, but we're not actually sure about that. So even with a ServiceError, we would still need another RawError if the responded error type couldn't be serialized to ServiceError.

[feiskyer]

/lgtm
/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: feiskyer, marwanad

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [feiskyer]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment