Implement error handling in case Composer timeouts or something else goes wrong

[Serializator]

Below a workflow run that failed because the Magento repository (repo.magento.com) was unresponsive.
https://github.com/weprovide/magento-docker-image/runs/2379055122

Right now this results in a corrupted strategy for the next step, which does not immediately indicate what went wrong.