Draft: Use the built-in GitLab CI cache mechanism

Retrieving the published data can be complicated on downstream deployment where the data is auth protected.

Since GitLab already provides a caching mechanism, use it to side-step the issue.

As a tradeoff, this means that data is cached per-branch, so new branches will not fetch any cached data. On the other hand this can be good since we will not have to deal with compatibility issues in the cached data when introducing incompatible changes in the data model.