Remove need to login with azcopy

[johnbradley]

The current Azure Blob Storage code requires that the user previously logged in with the azcopy command for uploading and downloading. Remove this requirement to avoid users needing to login multiple times.

We currently reference the bucket(container)/path with a plain URL:

DukeDSClient/ddsc/azure/api.py

Line 163 in 0b4eb27

destination_url = self.get_url(f"{project_path}/{directory_name}")

DukeDSClient/ddsc/azure/api.py

Line 173 in 0b4eb27

source_url = self.get_url(project_path)

If we used a SAS URL to reference the bucket(container)/path there would be no need to login with azcopy.

DukeDSClient/ddsc/azure/api.py

Lines 135 to 139 in 0b4eb27

	def get_sas_url(self, path, hours=6):
	account_name = self.service.account_name
	bucket_name = self.file_system.file_system_name
	token = self.get_sas_token(hours=hours)
	return f"https://{account_name}.blob.core.windows.net/{bucket_name}/{path}?{token}"

We are not currently doing so because the test environment does not currently allow creating SAS URLs.

[johnbradley]

Azure Storage Explorer seems uses different methods to call azcopy.
Here is what it does for downloading a file when SAS is not possible:

export AZCOPY_CRED_TYPE=OAuthToken;
./azcopy login;
./azcopy copy "..non_SAS_URL.../file1.txt" "/tmp/file1.txt" --overwrite=prompt --check-md5 FailIfDifferent --from-to=BlobLocal --recursive --log-level=INFO;
./azcopy logout;
unset AZCOPY_CRED_TYPE;

When creating a SAS URL is possible Azure Storage Explorer creates SAS URLs.