RFE: (dlrnapi-promoter) Better handle Error removing image docker.io/tripleoqueens/centos-binary-nova-placement-api:current-tripleo - UnixHTTPConnectionPool(host='localhost', port=None): Read timed out. (read timeout=60)

[Matt Young]

A recent queens promotion log looks like this:

```
TASK [Delete the local images from docker hub]

<snip/>

failed: [localhost] (item=[u'nova-placement-api', u'current-tripleo']) => {"changed": false, "item": ["nova-placement-api", "current-tripleo"], "msg": "Error removing image docker.io/tripleoqueens/centos-binary-nova-placement-api:current-tripleo - UnixHTTPConnectionPool(host='localhost', port=None): Read timed out. (read timeout=60)"}
```

This is happening here:

https://github.com/rdo-infra/ci-config/blob/master/ci-scripts/container-push/container-push.yml#L170

We're getting a read timeout, and failing the entire promotion. In this case the image is actually deleted.

there's a few ways to handle this...

- We can not fail if we can't delete something. Allow promotion to continue (in this case to report results, etc).

- (better) We can leave images in the repository, then age them out intentionally. this will allow partial failure, and 10 mins later when we attempt to re-promote, we don't need to pay the > 1 hour tax of pulling ALL the container layers from rdoproject, and pushing ALL the layers to docker.io.

[Matt Young (halcyondude)]

The full snip of the dlrnapi promoter script