Force re-auth if cached credentials are invalid

[ccapurso]

Overview

Attempting to run hcp connect will fail if a prior successful connection has been made but the associated credentials have expired or been invalidated. The underlying issue is rooted in how hcp-sdk-go caches credentials. An attempt to list HCP organizations using invalid credentials will fail with a 401 response. The following error is provided to the user:

Failed to get HCP organization information: [GET /resource-manager/2019-12-10/organizations][401] OrganizationService_List default &{Code:0 Details:[] Message:}

This issue can currently be resolved by running hcp disconnect which will perform a logout within hcp-sdk-go. This PR introduces a change to validate the obtained credentials using the IAM service GetCallerIdentity API endpoint. A logout (similar to hcp disconnect) is performed which will clear the credentials cache on disk in ~/.config/hcp/creds-cache.json. A re-authentication is initiated once the logout succeeds.

IAM Service mocks were added for testing which accounts for a large portion of the diff.

This PR also introduces a couple unrelated changes:

• A slight refactor such that the client members of HCPConnectCommand are no longer passed as method arguments and are instead accessed from the receiver
• Improved the error message provided when a cluster identified by -cluster-id does not exist

Related Issues/Pull Requests

This will be incorporated with a Vault PR once approved and merged

[VioletHynes]

Looks good to me! A few small comments. Note though that it's harder for me to speak for the veracity of the business logic as I'm less familiar with the subject matter.