WIP: AdaptiveExpectations : Usual/Strange regimes

[sbenthall]

As per #60

Introduces AdaptiveExpectations class which extends the FinanceModel with:

• 'usual' expectations
• when 'risky_expectations()' is called to inform an agent, it stochastically tests whether the finance model expectations are sufficiently distant (via Kolmogorov-Smirnov test) from the Usual expectations. If not, usual expectations are transmitted instead of the FinanceModel expectations.
• A parameter zeta that thresholds how different the FinanceModel predictions must be to break from usual.

This now can be used with the run_any_simulation script with the command line argument --expectations AdaptiveExpectations. But:

• I have not yet exposed the zeta parameter properly to that script.
• Nor have I made it so the simulation calibrates the usual expectations to the dividend process.

So this needs some more work, including testing, before merging.

[sbenthall]

This is getting there.

Next, I need help from @mesalas : What formula are you using for the institutional investors' derivation of their price/dividend ratio? It is important that the 'USUAL' expectations of the macro agents reflect this ratio. I'm wondering what's the best point to compute or configure this value.