Implement a (hidden) Markov model

[pawel-czyz]

Use TFP on JAX to define a new BMM:

X1 – X2 – ... – Xn 
 |   |          |
Y1   Y2 – ... –  Yn

However, this requires first fixing #161, so that the joint distribution returns tuples of arrays for $(X, Y)$, rather than sampling them jointly and unwrapping.