Using HTCondor with my Yadage Re-Cast --> REANA workflow

This is somewhat related to this post - How do I translate my recast workflow to a REANA workflow? - but I think a more advanced topic.

I would eventually like to submit my workflow to REANA and have it flock to HTCondor. There is indeed a nice example of flocking workflows to HTCondor from the REANA team - - but this is, unfortunately, written in serial, not yadage.

It seems that the key entry in the spec is that in each step of the workflow, one dictates that you should use a specific compute backend, namely compute_backend: htcondorcern. This is fine, but it would seem that it therefore requires that I modify by underlying workflow.yml which I have written for my recast workflow. Is this true?

This seems like it is less than ideal as I cannot easily transition back and forth between running my workflow with recast or with REANA. I will always have to modify the workflow itself. Or perhaps I should be passing this as a parameter to the workflow?