As can be seen in previous discussions - the general consensus is that failing early should be the default. I fully agree to this, but when using jobs for things like both linting and testing and even coverage; it makes no sense for those three to not be run independently disregarding outcome of either. One can obviously violate the “exit code” strategy and always return success to proceed in the pipeline, but that’s as you probably agree; a violation.
My suggestion would be a configurable setting in config.yml or even per job item. Seeing how I haven’t had the chance to test out workflows just yet I’m suspecting that the solution might be there already.