JSpecify: handling the return of a diamond operator anonymous object method caller

[akulk022]

In reference to the exception mentioned in the discussion for #791 .

Adding the test case to reproduce the same issue:

class Test {
    static class B<T>{
        String build(){return "x";}
    }
    static String testNegative() {
        //We were getting the aforementioned mentioned exception when we tried to do this
        return new B<>().build();
    }
}

All unit tests have passed for the changes that were made for this.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (0141aca) 86.93% compared to head (35c0cc3) 86.95%.

Additional details and impacted files

@@             Coverage Diff              @@
##             master     #858      +/-   ##
============================================
+ Coverage     86.93%   86.95%   +0.01%     
- Complexity     1888     1890       +2     
============================================
  Files            74       74              
  Lines          6215     6216       +1     
  Branches       1208     1209       +1     
============================================
+ Hits           5403     5405       +2     
  Misses          405      405              
+ Partials        407      406       -1     

Files	Coverage Δ
...rc/main/java/com/uber/nullaway/GenericsChecks.java	90.31% <100.00%> (+0.02%)	⬆️

... and 1 file with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[msridhar]

Instead of adding all the code above, why not just store the result of this getTreeType() call in a local variable, and if it is null, return Nullness.NONNULL? That is more general I think.

[akulk022]

I agree, I added it in the same condition, does this work?

[msridhar]

Better to do it after that condition and store the result of getTreeType() in a local variable, so we don't call it twice. Otherwise the logic looks good.