ENH: distinguish dimension of exogenous state space and number of exogenous variables

[albop]

So far, we silently make the assumption that they are both the same except in the case of i.i.d. shocks (where an empty grid is associated with exogenous variables).
In principle, the exogenous state-space could be bigger, as in heterogeneous agents models, or smaller for some families of exogenous shocks.
There are several design issues:

• there should be a gdp object distinguishing state-space and variables, that would be seamlessly compatible with our solution routines (modulo some adjustments)
• the distinction probably needs to extend to all processes too, as we also need to simulate the process (not just used its discretized form for solutions)
• when the state-space is smaller than the number of exogenous symbols, can we detect its size reliably
  ? (not sure at all)