Support for multivariate distributions

[dnclrhs]

Not sure if I misunderstand something.
Appears that UncertainFunction is unable to handle multivariate distributions.

Tried running

x = uv(scipy.stats.multinomial(n=1,p=[p1,p2,p3]))

and got the error

AssertionError: Input must be a  distribution from the scipy.stats module.

Which it clearly is.
Any idea?

[dnclrhs]

I notice

assert hasattr(
            rv, "dist"
        ), "Input must be a  distribution from the scipy.stats module."
        self.rv = rv

in mcerp, but the multinomial distribution of scipy has a _dist attribute, not a dist