Improve the test coverage of our built docker images
We currently don't build all of the docker images as part of the test phase. Instead, we build some only when we have merged to master
/live/test-cara
. As a result, we don't know if the builds are successful, nor do we know if the resulting images have the desired behaviour. Instead we should:
- Always build all images
- Push the images to the GitLab image repository with an appropriate
dev-${PIPELINE_ID}
tag - Test the pushed images
- (on deploy) Promote the pushed images, and never rebuild them at this stage