Through a combination of math and experimentation, I've found the correct computation of "human wealth" in the presence of risky returns. I've implemented it for RiskyAssetModel and FixedShareModel. On the next commit I'll get it working for the PortfolioChoice models, and then turn to ConsMarkovModel.

Seems to mostly work, *except* that MPCmin is now off by 0.00001 or so. This only becomes evident if you have an enormous top gridpoint, like m=10000. I have a guess why.

The proper human wealth and lower bound of MPC calculations are now in the "advanced" portfolio solver in ConsPortfolioModel, but not actually used in the consumption function. Calculation of the limiting share has been tweaked to be more accurate.

The "basic" solver in ConsRiskyAssetModel now goes through a loop to refine its search for optimal portfolio share at each aNrm gridpoint by "zooming in". We will probably want to add a parameter to specify how many times to do this loop, but it's currently set to 3.

I broke everything

Incorporate proper discounting for human wealth into ConsMarkovModel so that extrapolation works properly.