Fiscal updates

[jdebacker]

This PR will make some updates to how fiscal related variables are computed in OG-USA. The following changes will be made:

1. Fix an indexing error in the calculation of foreign debt holdings, D_f.
2. Move the lines of code to determine the path of foreign and domestic debt holdings into the fiscal.get_D_G_path() function.
3. Create a function to compute transfers within fiscal.py (this and the above help to remove conditional statements related to the balanced budget option from the main TPI (and SS) code.
4. Allow non-zero government spending in the balanced budget configuration.

[codecov-io]

Codecov Report

Merging #563 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master     #563   +/-   ##
=======================================
  Coverage   84.43%   84.43%           
=======================================
  Files          47       47           
  Lines        6613     6613           
=======================================
  Hits         5584     5584           
  Misses       1029     1029           

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3fd31f8...3fd31f8. Read the comment docs.

[jdebacker]

@rickecon This PR is ready for review.

[rickecon]

@jdebacker . The tests timed out on the Travis CI testing for Python 3.7. The last test to pass in the 3.7 CI unit tests was:

ogusa/tests/test_get_micro_data.py::test_get_data[Baseline] Running current law policy baseline
Running current law policy baseline
Year:  2029
Year:  2030

Right after that, the CI testing started giving Dask worker memory warnings like the following,

distributed.worker - WARNING - Memory use is high but worker has no data to store to disk.  Perhaps some other process is leaking memory?  Process memory: 3.12 GB -- Worker memory limit: 4.18 GB

and the Python 3.7 testing stopped with the following message.

The job exceeded the maximum time limit for jobs, and has been terminated.

This is a Dask parallelism issue. As is noted in Issue #562, we should probably calculate the total amount of RAM available to the Travis CI cores, divide that by the number of cores, and then try to stay under 60% of that on each core.

[jdebacker]

@rickecon I restarted the test for 3.7.

I'm not too worried about the dask issue on Travis CI. It's simple enough to try the test again for those instances when a worker runs into an issue. This is about the 3rd time this has happened to me in the last 2 weeks and I've always has success on the second attempt.

If we are hitting the time limit, we can flag some additional tests to not run on Travis CI.