What I've found most useful about writing Puppet recipes is it forces me to think about why something is configured the way it is. I'm not really certain Puppet would give me the same advantage if I were just using it clone existing machines (where I have no idea why things were setup the way they were).
I will follow this discussion and I'm happy to answer to your remarks/questions.
First point, we think that this kind of tools can makes you win a lot of time writing your own manifests. In a second hand, it could help for migrations of non-production services and even analyse towards your configuration (and provide an usable output). There will be other usages, still under development, that we will be happy to present you when it will be done ;)
Second point, this is true, Pysa and Blueprint are similar in the idea, as it is said in the documentation. However, it doesn't provide the same result (and doesn't have the same goals). A quick comparison has been written in the documentation (https://pypi.python.org/pypi/Pysa#notes).
Don't hesitate to post any other remark, question or requirement.
Also, how does this differ from BluePrint? http://devstructure.com/blueprint/