WIP implementation of standard API functions

[corakingdon]

@davidanthoff Questions:

• What should happen if the user calls compute_scc for a year not in the model's time index? the index is 2005:10:2595. Currently if they call compute_scc(year=2020) this will error and say they need to provide a valid year. Do we want to implement interpolation so that if they request a 2020 SCC value, it would interpolate between a 2015 SCC value and a 2025 SCC value? this would cause problems for the compute_scc_mm function since more than one marginalmodel is used.
• Do we want to implement Ramsey discounting option? Currently the only available discounting keyword argument is prtp and constant discounting is used.
• Since there's currently no ramsey discounting, I made the default prtp value be 0.03. Let me know if I should change this.
• How exactly do we want to calculate the SCC since marginal damages are on 10 year timesteps? Currently I implemented the simplest way of just discounting each 10yr value then multiplying each value by ten. Alternatively, we could interpolate (or do a step function) to have annual marginal damages, then discount each annual marginal damage value.

And I still need to update the README

[davidanthoff]

I think throwing an error for non-existing years is the right way.

I think we should compute the SCC differently: we should simply take the difference between welfare.UTILITY between the two runs. That will give us the SCC in welfare units, so we then still need to normalize with marginal welfare of the period of the emission pulse. Should we merge this PR as is, and then I make the changes to use that method, and you can use it as a template for the other models from that family?