Since the definitions and pipeline sections are separate, how can we have two docker steps that use different amounts of memory?
In my case, I do all the building in docker. So for the dev branches, I run docker build to do the basics setup and unit tests. But in the deployment branches, I also do the full builds. This means the deployment builds are done in `size: 2x` steps while the dev builds are done in normal size steps.
But I need to run the dockers those with different service limits on memory also so that docker can use the increased memory while building.
I can make all the docker steps 2x, but then I'm wasting memory and money for steps that don't need the extra memory.
