Add test that shows bug in GainsTax function and fix bug

[martinholmer]

This pull request adds a test to illustrate a bug in the GainsTax function and then fixes that bug. This bug was reported by @codykallen in issue #977. The fix for this bug in this pull request is logically equivalent to the code revision suggested by @codykallen in pull request #979, but eliminates the duplication of code inherent in #979. Thank you, @codykallen, for your helpful bug report in #977 and your correct diagnosis of the bug's cause in #979.

As an aside for other Tax-Calculator contributors, the bug report in #977 is an example to be emulated: it describes the problem and provides a simple test to demonstrate the problem. #979 goes the extra mile by actually suggesting how to fix the bug reported in #977.

What I have done in this pull request is two things:
(1) created a new unit test based on the bug report in #977, and
(2) fixed the bug (that is, revised the code so that the new test does not fail).

@MattHJensen @feenberg @Amy-Xu @GoFroggyRun @andersonfrailey @zrisher

[codecov-io]

Current coverage is 98.27% (diff: 100%)

Merging #981 into master will not change coverage

@@             master       #981   diff @@
==========================================
  Files            35         35          
  Lines          2085       2085          
  Methods           0          0          
  Messages          0          0          
  Branches          0          0          
==========================================
  Hits           2049       2049          
  Misses           36         36          
  Partials          0          0          

Powered by Codecov. Last update 9b67e16...e4618b6

[martinholmer]

Before the #981 bug-fix code change in the functions.py file, the results of the new test_Calculator_mtr_when_PT_rates_differ test were as follows:

>       assert np.allclose(mtr1, mtr2, rtol=0.0, atol=1e-06)
E       assert <function allclose at 0x102f93410>(array([ 0.188]), array([ 0.438]), rtol=0.0, atol=1e-06)
E        +  where <function allclose at 0x102f93410> = np.allclose

So, before the bug fix, the marginal tax rate on LTCG (p23250) was different before and after the reform that made pass-through tax rates different from tax rates on non-pass-through income: instead of being the same, the mtr on LTCG was 0.188 (the 15% CG rate plus the 3.8% NIIT rate) when II and PT tax rates were identical and 0.438 (the 40% II rate plus the 3.8% NIIT rate) after the reform that made all II rates equal to 40% and all PT rates equal to 30%.

After the bug fix in #981, the mtr on LTCG is 0.188 both before and after the reform that makes II and PT rates different.

@MattHJensen @feenberg @Amy-Xu @GoFroggyRun @andersonfrailey @zrisher

[MattHJensen]

@martinholmer, this looks good to me. @codykallen, thanks a lot for the bug report and suggested fix!

@martinholmer, do you think it is ok to merge this now? I think it deserves a TC version bump and redeploy to TaxBrain asap given that I know there are currently users on TB who would be affected.

cc @talumbau

[martinholmer]

@MattHJensen said:

@martinholmer, this [#981] looks good to me. @codykallen, thanks a lot for the bug report and suggested fix!

@martinholmer, do you think it is ok to merge this now? I think it deserves a TC version bump and redeploy to TaxBrain asap given that I know there are currently users on TB who would be affected.

Yes, I'm comfortable with the changes in pull request #981, but was hoping to get a thumbs up from @codykallen before doing the merge. But I haven't heard anything from his this morning.

[martinholmer]

Merged pull request #981 resolves issue #977 and incorporates the bug-fix strategy (but not tactics) of pull request #979.