You by definition start with a postulate that there is an "unwanted mental health effect" you want to reduce. We'll assume for the moment it isn't permanent schizophrenia. To see if you've reduced it, you need a control group getting the "wild type", who will be at risk of getting permanent schizophrenia to the best of our current knowledge. If that risk is real, there's also a real chance the causative agent(s) will be administered to the control group.
BTW, two of the major active ingredients are already legal drugs or being tested in the US:
https://en.wikipedia.org/wiki/Tetrahydrocannabinol#Marinol (Schedule III!)
https://en.wikipedia.org/wiki/Nabilone
https://en.wikipedia.org/wiki/Nabiximols (in US Phase III drug trials)
So your initial premise that the US government hasn't been allowing medical research is not quite the case. However, unlike recreational use, pharmaceutical use has a much higher tolerance of side effects, there's medical gain to be balanced against risks, or so you believe and keep testing at each stage of the drug trials.
On the other hand, with the recent state-wide decriminalizations we're "conducting" a rather while uncontrolled experiment right now that ought to be able to yield better data on the biggest concerns, although that sort of epidemiological research is really, really hard, and few are unbiased.