SimpleParamService doesn't filter estimator list by problem if dataset large and "all" estimators requested

[TimoFlesch]

The following code block is intended to filter the list of all implemented estimators by the problem (backdoor or iv). It doesn't do this if the dataset exceeds 1000 rows:

        if patterns == "all":
            if data_rows <= 1000:
                return [e for e in self.estimator_names if problem_match(e, problem)]
            else:
                warnings.warn(
                    "Excluding OrthoForests as they can have problems with large datasets"
                )
                return [e for e in self.estimator_names if "OrthoForest" not in e]

As example, compare the following output: 1/ ac.cfg.estimator_names_from_patterns("backdoor","all",100) yields:

['backdoor.auto_causality.models.Dummy',
 'backdoor.auto_causality.models.NewDummy',
 'backdoor.econml.metalearners.SLearner',
 'backdoor.econml.metalearners.TLearner',
 'backdoor.econml.metalearners.XLearner',
 'backdoor.econml.metalearners.DomainAdaptationLearner',
 'backdoor.econml.dr.ForestDRLearner',
 'backdoor.econml.dr.LinearDRLearner',
 'backdoor.econml.dr.SparseLinearDRLearner',
 'backdoor.econml.dml.LinearDML',
 'backdoor.econml.dml.SparseLinearDML',
 'backdoor.econml.dml.CausalForestDML',
 'backdoor.auto_causality.models.TransformedOutcome',
 'backdoor.econml.orf.DROrthoForest',
 'backdoor.econml.orf.DMLOrthoForest']

2/ ac.cfg.estimator_names_from_patterns("backdoor","all",1001) yields:

['backdoor.auto_causality.models.Dummy',
 'backdoor.auto_causality.models.NewDummy',
 'backdoor.econml.metalearners.SLearner',
 'backdoor.econml.metalearners.TLearner',
 'backdoor.econml.metalearners.XLearner',
 'backdoor.econml.metalearners.DomainAdaptationLearner',
 'backdoor.econml.dr.ForestDRLearner',
 'backdoor.econml.dr.LinearDRLearner',
 'backdoor.econml.dr.SparseLinearDRLearner',
 'backdoor.econml.dml.LinearDML',
 'backdoor.econml.dml.SparseLinearDML',
 'backdoor.econml.dml.CausalForestDML',
 'backdoor.auto_causality.models.TransformedOutcome',
 'iv.econml.iv.dr.LinearDRIV',
 'iv.econml.iv.dml.OrthoIV',
 'iv.econml.iv.dml.DMLIV',
 'iv.econml.iv.dr.SparseLinearDRIV',
 'iv.econml.iv.dr.LinearIntentToTreatDRIV']