Thanks, I just had the same thought, looking at the thread you had linked to. Thanks for your reply!
But more generally speaking, is it correct to say that the difference between webPPL and pymc with regards to this example (or perhaps in general), that webPPL supports the (simple) inference algorithms of exact enumeration and rejection sampling, whereas pymc does not directly support this, so we have to solve this example either with manual rejection sampling or by using the construction with Categorical variables and the lookup table?