Resolve ADLS2 Mount Diff and Incorrect error messages

[lawrencegripper]

Fixes #37 #38

Changes:

• Add additional try-except so unmount error isn't returned in place of actual error on mount failure [ISSUE] Azure Mount ADAL throws wrong error hiding error details #37
• Remove requirement for / in directory name as mounting the root of the Datalake is a valid action (highlighted by the tests)
• Remove the additional / in the python mount command which caused a diff to be detected [ISSUE] Azure ADAL Mount always detects diff due to extra slash #38
• Add a repeatable integration test for Azure Resources, initially implementing this resource [TEST] Acceptance tests for Azure Mounts #20

The individual fixes for the bugs are relatively simple, majority of the work has gone into getting a repeatable end2end acceptance test setup for this Azure resource (and hopefully to use for all Azure integration tests).

Integration testing is done by running ./integration-environment-azure/run.sh:
1.This creates prereq from scratch (workspace, storage etc) in azure
2. These are output from the Terraform prereq.tf into environment variables
3. Any go tests starting with TestAccAzure are invoked using this envs
4. Script cleans up the environment

Note: We create each environment from scratch to ensure no state from previous runs breaks tests and full end2end on new workspace is successful. When testing SKIP_CLEANUP can be used to avoid recreating the env each time. More details in the README.MD file.

Would be great to get this script run as part of travis like #45 - are you happy for us to hook it up?

This work is co-authored with @stuartleeks

The tests seem to block on #33 occasionally when creating the cluster intermittent failures occur as seen before related to associated worker environments.

[stuartleeks]

This PR ended up being a little bigger than expected as we needed to set up some infra in the code around creating the additional Azure resources needed to integration test these changes. Hopefully, this will make it easier for further tests of this nature in future :-)

[stikkireddy]

hey @lawrencegripper & @stuartleeks thanks for the PR I should be able to finish reviewing this by the afternoon today (EST)

[stuartleeks]

@stikkireddy - awesome, thanks for the update :-)

[stuartleeks]

Hey @stikkireddy - any update on this? Happy to answer any questions you may have 😃

[stikkireddy]

LGTM @stuartleeks this still has issues with the tests sporadically and I believe the solution for making sure that the clusters api works properly after creating a workspace is proposed to be solved via #53. I will merge this in and hopefully that #53 can help successfully run this test 10/10 times successfully.