Consider moving snappass-test image to GHCR

[dimaqq]

From stand-up: Juju 4.0.1 was released, Tony tried to run integration Jubilant integration tests.
Some got stuck because Docker Hub throttling kicked in. This could be because snappass-test needs Redis which is pulled from Docker Hub.

If we pushed the required resource image to GHCR (or quay.io or our infra), we'd avoid this problem.

Aside: I find it a bit weird that redis-image details are visible on charmhub...

[dimaqq]

Charmhub reference for the redis image is: https://api.charmhub.io/api/v1/resources/download/charm_rQFqM8IGBua8n8rSi3n7PLukIdoRz7Ie.redis-image_3

And same for snapppass-image is: https://api.charmhub.io/api/v1/resources/download/charm_rQFqM8IGBua8n8rSi3n7PLukIdoRz7Ie.snappass-image_3

There are no declared upstream resources for this charm.
(For comparison, traefik-k8s declares docker.io/ubuntu/traefik:2-22.04)

I think the issue may be that the integration test deploys the charm without explicit resources:

juju.deploy("snappass-test")

The charm readme contains this example:

$ juju deploy ./snappass-test/snappass-test_ubuntu-22.04-amd64.charm \
    --resource snappass-image=benhoyt/snappass-test \
    --resource redis-image=redis

I'm actually a little lost whether perhaps we're mixing charmhub references (our tests) and explicit docker hub references (charm readme)...