Fix sharing update

[dlpzx]

Feature or Bugfix

• Bugfix

Detail

When we import a dataset stack, if the S3 location was already registered data.all does not create a storage location. The issue is that for datasets where data.all needs to create a storage location:

1. the first time that it creates the stack it detects that there is no storage location and it creates the corresponding CFN resource
2. the first time that it UPDATES the stack it detects the storage location from 1. and it deletes the CFN resource
3. the next time that it UPDATES the stack it does not detect any storage location (it was deleted in 2.) and it creates the CFN resource again.

To fix this behavior, in V1.5 we will use Lambda custom resource to check the storage location and avoid CFN resources. But for previous versions, this PR includes:

• in the method that checks the existence of an storage location, we filter by the roleArn of the location. If the roleArn is the dataallPivotRole then we assume that it was created by the dataset, which means that existing_storage_location = False

I tested locally but with actual stacks being created. No additional policies are needed

Relates

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[noah-paige]

Overall this looks good to me and a very quick fix!

Only note would be to add some additional comments to be more verbose on what we are doing here (it is a little confusing at first glance seeing we return False on both errors and if data all registers location)