Consider Post-Build Testing Requirements

[CodeGat]

A probable output to the Entrypoint workflows would be something like a link to a singularity container with the given model component installed. But model components need more testing than that, and this testing could be quite varied between model components.

Things to Consider

• Would there be a common core of tests that need to be run? For example, testing if the model component can be installed as a dependent to multiple different models.
• Should we just return the image, in which case:
  • Would we return a reference to a singularity image with the model installed? Similar to a [docker commit](https://stackoverflow.com/questions/62698416/what-is-the-equivalent-of-docker-commit-in-singularity) image?
  • Or a reference to a running container on GHA? How would this work for private repository minutes?
  • Or a way of the user to build the .sif themselves? Must remember the mount points/etc.

[CodeGat]

Rather than a pytest wrapper for commonality, maybe (also?) support running post-build scripts under .github/build-ci/test-scripts, or something, that is executed in the container. Might be easier for model devs rather than having to call pytest to call a script.