Policy Generator tool, first pass

[ahouseholder]

This PR contains a python class that generates an SSVC decision policy (aka a "tree") that fully enumerates a set of decision points and their values, and assigns an outcome to each input value tuple. It does so by

• constructing a partial order over all the input value tuples
• generating a topological sort over the resulting partial order
• assigning outcomes according to a set of provided weights
• emitting a CSV that replicates the format we've previously used to represent decision policies
• (one column for each decision point, plus an outcome column filled in with the assignments made)

If you run policy_generator.py directly, it outputs a policy that's rather close to the Deployer 2.1 tree.

At this point, my goal is to get the module in place with this PR.

We will need to figure out what the UI is for it, and how to make it easier to use. Right now, you have to write python code to use it, an example of which is provided in ssvc.policy_generator.main, but it can probably be made easier to use. One usability issue I can see coming soon is how someone figures out which decision point (and version) to use as input, and how they represent their outcome set. My preference would be to create a separate issue to come back to the CLI side of this tool in the future. For now, I think it's usable enough for someone comfortable with some light coding-by-example, such as a Jupyter notebook user. I created a separate issue to capture the UI followup task (#366) .

One notable addition in this PR is the creation of src/ssvc/outcomes/base.py, which contains a python representation of an OutcomeGroup and OutcomeValues, and src/ssvc/outcomes/groups.py which provides a variety of examples thereof. I think we discussed not versioning outcome sets (I called them outcome groups which have outcome values here but if there's a preference to rename them simply outcomes with outcome values, that's an easy change to make.)

This PR

• resolves Tool to distribute outcome set across a decision point group #336
• resolves MoSCoW outcome set #337
• partially addresses (by modeling the outcome set only) Eisenhower matrix tutorial #335

I included #337 and #335 because they were good exercise for building OutcomeGroup objects.

[j---]

This looks like it has all the parts it needs. I didn't do independent runtime testing.