Adds a non-OO solve_one_period function for the ConsPerfectForesight model. Will do for the other solvers as well.

The value function in the PF model isn't right when there's an artificial borrowing constraint (in current HARK). This commit makes it less wrong, but I don't think it's *right* yet.

Still need to add CubicBool and vFuncBool options.

This ended up 75 lines longer on net, but it's much easier to understand.

Had an off-by-one error in the "coefficient hacking" line, creating a discrepancy in the fourth decimal place. Now fixed,

Matches behavior exactly for the default dictionary. Need to add vFunc functionality, then expand to discretized version.

This commit has the new solver commented out in the type init method so as to not create a ton of new failing tests while it's still a WIP.

Have not tested fixed share functionality, but I expect it works.

Still doesn't handle non-independent shock distributions.

This is where we really start to see that subfunctions for solvers really should be used, as this is almost entirely a copy-paste from ConsIndShock. The idea behind the OO solver encapsulation wasn't bad, but the resulting code was horrible.

This solver also makes clear that we need encapsulation within solution methods. That will happen after I've "flattened" everything.

Whoops.

Value function capability has not been set up yet, will do on next commit. The Markov model code is where we see most clearly that the "timing change" will make everything better and easier to read. There's a *lot* of confusing stuff going on due to the fact that both t and t+1 activities are being handled by the period-t solver. Ugly.

Forgot to test CubicBool=True on prior commit, but there was only one tiny error, 3 character fix. Began to implement value function, but ran out of time.

Also fixed a bug/error in the existing vFunc code that made it incapable of actually solving. The good news is that the new code exactly replicates the old code. The bad news is that something is wrong with the vFunc extrapolation code, identically in both versions. Shouldn't be too hard to fix.

The math for the limiting linear function for the pseudo-inverse value function must not be correct, as that's what was causing the vFunc jump. It's been commented out for now.

Need to look at whether the cFunc extrapolation code is correct for ConsMarkovModel.

This was very easy, basically a copy-paste of IndShock's simple solver with 3 lines added. Removed all the cubic and vFunc stuff to match current capability.

I kept the vFunc and DiscreteShare capability in this solver, and looks like I could/should have left vFunc and CubicBool in the other one. Also found and fixed two bugs/typos in the existing code.

I've tested it on the PersistentIncomeShock model, and it reproduces the solution exactly. Strangely, it's 5-6x *faster* than the existing code, even though I made this version by "flattening" the old version. That probably means there's a hidden loop of some kind that I (unknowingly) skipped over.

Also: I think PersistentShockType doesn't initialize properly with default parameters. I'll look into it.

This version *only* has the non-independent distributions code, which turns out to be easier to adapt (when starting from the ConsIndShock solver). I will add the independent distributions version shortly. I also added cubic spline capability, but I'm slightly worried there's a small bug in it.

Strangely, my "flat" code is again *much* faster than the current code: 10-12x faster. I have absolutely no idea why. The new non-independent solver is faster than the old independent solver.

ALSO: The old independent solver doesn't generate the same solution as the non-independent solver. Some discrepancy is to be expected, but it's bigger than I expected.

It was trying to use broken values before.

RiskyAsset simple solver can now handle IndepDstnBool=True. Solution is internally consistent with False version. Existing solver has a discrepancy between solutions.

There's also a problem with independent distributions and CubicBool=True, but I'm out of work time today.

There was an errant point at aNrm=0, just had to remove it.

The "medical shock" model now has a simple solver that replicates behavior of the existing solver. The current OO version's CubicBool capability is broken; it works in the new simple solver, but I haven't fixed it in the OO code.

It looks like the tests for this model just make sure that it can run and simulate, but I've verified that the solution reproduces exactly on my own. As with other models, this solver runs faster than before, but not by as much: 0.5-2.5x faster.

There are improvements and revisions to this model that should happen, but they're for another branch.

I had cut this on a prior commit, but now that I've made other (small) improvements to other models, it seems fine to include it. It's almost entirely a copy-paste from ConsIndShock, as you would expect.

Last bit of functionality has been added to the simple solver for ConsPortfolioModel: non-independent income and return shocks. Value function differs slightly from legacy code, but that's because the old code does not construct end of period value function correctly (doesn't use CubicInterp).

These updates include previously merged branches that didn't get entries. Whoops.

Had to move one very small agent type into its own file. It looks like it's a PortfolioConsumerType with a slightly different solver.

I said that this was like untangling rope, but it's more like brushing a badly groomed dog.

Redirect to another file.

This is now handled by the same solver: just set boolean flag and let it run.

There used to be a special solver in ConsRiskyAssetModel.py that did the "basic" portfolio model, without discrete choice nor adjustment frictions. A simple solver of that kind has now been added.

There is now a "simple solver" for a model in which the risky share is *fixed* (but might vary by age), but is not necessarily 0 or 1. This previously existed, was temporarily exiled to a new file, and now has been restored.

Two very small compatibility changes