Previously there were a couple models which used different owner ids by
default, which could make tests relying on them fail if someone wasn't
expecting that. This change ensures a uniform owner id between
resources.
This commit adds a test for a case where an EBS volume has no tags. When
an EBS volume has no tags, calls to the aws ec2 endpoints `create_volume`
and `describe_volumes` do not include the `Tags` key in the
`response.Volumes[]` object.
However, moto does include the `Tags` key in this case. This discrepancy
in behaviour can result in code passing a moto test but failing in
production.
Sample snippets that trigger this condition:
```
def create_volume_and_then_get_tags_from_response():
client = boto3.client('ec2', region_name='us-east-1')
volume_response = client.create_volume(
Size=10,
AvailabilityZone='us-east-1a'
)
keys = volume_response['Keys']
```
```
def create_volume_and_then_get_tags_from_describe_volumes():
client = boto3.client('ec2', region_name='us-east-1')
volume_response = client.create_volume(
Size=10,
AvailabilityZone='us-east-1a'
)
volume_describe_response = client.describe_volumes()
keys = volume_describe_response['Volumes'][0]['Keys']
```
Both sample snippets will succeed in a moto test, but fail with a
`KeyError` when using the aws api.
The logic which contructed a list of values for parameters with
multiple values was flawed in that e.g. `Subnet.1` and `Subnet.10`
would be have their values counted against `Subnet.1` because they
share a prefix.
This now checks for a starting `.` before counting that name as
having the requested prefix.
* Adding owner-id/OwnerId to the AMI classes to allow the value to be specified to test filtering images based on owner.
* Added default AMIs and filtering by owner-id
* Fixed some tests
* Fixed more random tests
* Updated MANIFEST
* .
