Flaky test in nightlies: target distribution data check

[dsherry]

I saw the following fail in the linux nightlies yesterday

FAILED evalml/tests/data_checks_tests/test_target_distribution_data_check.py::test_target_distribution_data_check_unsupported_target_type[double]
...
        if target_type in ["integer", "double"]:
>           assert target_dist_check.validate(X, y) == {
                "warnings": [],
                "errors": [],
                "actions": [],
            }
E           AssertionError: assert {'actions': [...rning', ...}]} == {'actions': [...warnings': []}
E             Omitting 1 identical items, use -vv to show
E             Differing items:
E             {'actions': [{'code': 'TRANSFORM_TARGET', 'metadata': {'column': None, 'is_target': True, 'transformation_strategy': 'lognormal'}}]} != {'actions': []}
E             {'warnings': [{'code': 'TARGET_LOGNORMAL_DISTRIBUTION', 'data_check_name': 'TargetDistributionDataCheck', 'details': {'shapiro-statistic/pvalue': '0.969/0.018'}, 'level': 'warning', ...}]} != {'warnings': []}
E             Use -v to get the full diff

evalml/tests/data_checks_tests/test_target_distribution_data_check.py:59: AssertionError

Tagging @ParthivNaresh since I think he touched this code recently and may have an idea on the fix

[freddyaboulton]

I think we just got a flake on a different test for build_conda_pkg:

https://github.com/alteryx/evalml/pull/2585/checks?check_run_id=3242371966

[dsherry]

Got it @freddyaboulton , so it was a different test but from the same file / area.

[freddyaboulton]

Looks like we saw a similar flake due to rounding in the 3.9 nightlies: https://github.com/alteryx/evalml/runs/3503548207

Reopening this issue to track if it happens again!

[ParthivNaresh]

I have a fix for this, will be putting it up now